Carnotaurus sastrei Bonaparte,
THE Horned, Lightly Built Carnosaur
FROM THE Middle Cretaceous of Patagonia
J. F. Bonaparte, F. E. Novas, and R. A. Coria^
CONTENTS
ABSTRACT. 2
RESUMEN. 2
INTRODUCTION. 2
SYSTEMATICS AND PALEONTOLOGY. 2
Cranial Skeleton. 3
Skull. 3
Lower Jaws. 6
Hyoid Apparatus. 6
Comparisons of Skull and Lower Jaws. 6
Axial Skeleton. 10
Vertebral Column. 10
Ribs. 19
Gastralia . 21
Haemal Arches. 21
Appendicular Skeleton. 21
Pectoral Girdle and Forearm. 21
Pelvic Girdle and Hindlimb. 28
Skin. 32
General Comparisons. 32
Is Carnotaurus sastrei a Carnosaur?. 32
Comparison with Jurassic and Cretaceous Carnosaurs of South America. 33
Comparisons with Tyrannosauridae. 34
Comparison with Allosauridae. 35
Comparison with Ceratosauridae. 35
Comparison with Theropods from Egypt and India. 36
Summary. 38
DISCUSSION .. 39
ACKNOWLEDGMENTS. 40
LITERATURE CITED. 40
Gallardo 470, 1405 Buenos Aires, Argentina.
Contributions in Science, Number 416, pp. 1-42
Natural History Museum of Los Angeles County, 1990
ABSTRACT. Carnotaurus sastrei Bonaparte, 1985, is known from Cretaceous deposits referred to the
Gorro Frigio Formation (Albian or Cenomanian) of Chubut Province, Patagonia, Argentina. It is an
unusual theropod, with two short, stout frontal horns; a very weak contact between dentary and post¬
dentary bones; cervical vertebrae quite derived, with small neural spines and very pronounced epipophyses;
forelimbs extremely reduced; and long, slender hindlimbs. Carnotaurus sastrei shares several derived and
primitive characters with Abelisaurus comahuensis Bonaparte and Novas, 1985, from the Allen Formation,
Maastrichtian of Patagonia, Argentina: maxilla short and high in its anterior half, a small maxillary fenestra,
nasals strongly decorated, wide contact between postorbital and lachrymal above the orbit, elongate
quadrate, and ventral branch of the squamosal ventrally directed. These features suggest that both species
belong to the same family, Abelisauridae Bonaparte and Novas, 1985.
The Noasauridae, represented by Noasaurus kali Bonaparte and Powell, 1980, from the Lecho
Formation, Maastrichtian, northwestern Argentina, appears to be related to the Abelisauridae because
the contained species have a similar type of cervical vertebrae, i.e., with reduced neural spines and well-
developed, spine-like epipophyses. Comparative analyses suggest that Abelisauridae and Noasauridae share
more characters with Ceratosauridae than with any other Theropoda. Two superfamilies are recognized;
Ceratosauroidea (including Ceratosauridae, Abelisauridae, and Noasauridae) and Tyrannosauroidea (in¬
cluding Allosauridae and Tyrannosauridae).
RESUMEN. Carnotaurus sastrei Bonaparte, 1985, proveniente de depositos Cretacicos referidos a la
Formacion Gorro Frigio (Albiano o Cenomaniano) de la Provincia del Chubut, Patagonia, Argentina, es
un inusual teropodo con dos cortos y robustos cuernos frontales; mandibula con debil contacto entre
los elementos dentarios y postdentarios; vertebras cervicales especializadas, con espinas neurales pequehas
y epipofisis muy pronunciadas; miembros anteriores sumamente reducidos; y miembros posteriores largos
y graciles. Carnotaurus sastrei comparte una serie de caracteres derivados y primitivos con Abelisaurus
comahuensis Bonaparte y Novas, 1985, de la Formacion Allen, Maastrichtiano de Patagonia, Argentina:
maxilar corto y alto en su sector anterior, fenestra maxilar pequeha, nasales fuertemente decorados,
postorbital y lacrimal con amplio contacto por encima de la orbita, cuadrado largo, y escamoso con rama
ventral dirigida ventralmente. Estos caracteres sugieren que ambos taxones pertenecen a la familia Abe¬
lisauridae Bonaparte y Novas, 1985.
La familia Noasauridae, representada por Noasaurus kali Bonaparte y Powell, 1980, de la Formacion
Lecho, Maastrichtiano, NW de la Argentina, parece estar relacionado a la familia Abelisauridae, debido
a que presenta un tipo similar de vertebras cervicales, con espinas neurales reducidas y epipofisis desa-
rrolladas. Las comparaciones efectuadas sugieren que Abelisauridae y Noasauridae comparten mas ca¬
racteres con los Ceratosauridae que con otros teropodos conocidos. Se reconocen dos superfamilias:
Ceratosauroidea (que incluye a Ceratosauridae, Abelisauridae, y Noasauridae) y Tyrannosauroidea (que
incluye a Allosauridae y Tyrannosauridae).
INTRODUCTION
During fieldwork of the 8th Paleontological Ex¬
pedition to Patagonia, within the project “Jurassic
and Cretaceous Terrestrial Vertebrates of South
America” sponsored by the National Geographic
Society, the skeleton of a theropod of rather great
size was excavated. It was in a good state of pres¬
ervation, with all of the preserved bones perfectly
articulated. Skin impressions were found in the area
of its right side, perhaps the first recorded among
the carnosaur dinosaurs. Fiowever, weathering had
affected the central and distal parts of the tail, most
of the tibiae, both fibulae, and the hind feet. The
remainder of the skeleton was complete, articulat¬
ed, and with some lateromedial deformation that
was more pronounced in the skull.
Carnotaurus sastrei was briefly described by Bo¬
naparte (1985). It was noted that there are strong
differences between it and the Cretaceous carno-
saurs from the northern continents, especially in
the skull, axial skeleton, and in the striking reduc¬
tion of the forelimbs. Bonaparte and Novas (1985)
recognized that Abelisaurus comahuensis repre¬
sented a new family of the Carnosauria, Abelisaur¬
idae, to which C. sastrei was referred. This family
is clearly defined by characters of the skull: large
infratemporal fenestra; elongated quadrate; poste-
2 ■ Contributions in Science, Number 416
riorly directed squamosal with a ventral, rod-like
process; and a small maxillary fenestra located very
near the preorbital opening.
The marked anatomical differences of this family
of Patagonian theropods, in comparison with ther-
opods from the Northern Hemisphere, have been
interpreted by Bonaparte (1985, 1986b) and Bo¬
naparte and Novas (1985) to be the result of the
long geographic separation of the Laurasian and
Gondwanian supercontinents. According to the pa-
leogeographic evidence (Bonaparte, 1986b), the iso¬
lation occurred from the late Middle Jurassic to
the Late Cretaceous (Campanian).
The following abbreviations are used in this pa¬
per: AMNH, American Museum of Natural His¬
tory; MACN-CH, Museo Argentino de Ciencias
Naturales “B. Rivadavia,” Coleccion Chubut; and
USNM, United States National Museum of Natural
History.
SYSTEMATICS AND PALEONTOLOGY
Order Saurischia Seeley, 1888
Suborder Theropoda Marsh, 1881
Infraorder Carnosauria Huene, 1920
Family Abelisauridae
Bonaparte and Novas, 1985
Bonaparte, Novas, and Coria: Carnotaurus sastrei
Genus Carnotaurus Bonaparte, 1985
Type species Carnotaurus sastrei
Bonaparte, 1985
HOLOTYPE. MACN-CH 894. Almost com¬
plete skeleton, with skin impressions, lacking me¬
dial and distal parts of the tail, central and distal
parts of the tibiae, both fibulae, and hind feet.
LOCALITY. Estancia Pocho Sastre, near Bajada
Moreno, Department of Telsen, Province of Chu-
but, Patagonia, Argentina (Fig. 1).
HORIZON AND AGE. Near the top of the
Gorro Frigio Formation, Cretaceous (Albian or
Cenomanian).
REVISED GENERIC DIAGNOSIS. Abelisaurid
carnosaur, with skull shorter and higher than in
Abelisaurus and other theropods, and with deep
snout and prominent frontal horns. Orbits divided
into two parts: an upper, rounded section antero-
laterally projected for the eyes, with a pronounced
posterodorsal orbital wall, and a lower, dorsoven-
trally elongated section. Supratemporal opening
small, with parietal and squamosal forming a high
posterior wall and having a low, lateral border.
Infratemporal and preorbital openings smaller than
in Abelisaurus. Quadrate very high, and squamosal
having a short, rod-like ventral projection. Loose
contact between dentary and postdentary bones,
forming a large mandibular fenestra. Cervical ver¬
tebrae behind the axis with reduced neural spines
and high, well-developed epipophyses. Sacrum with
seven fused vertebrae. Forelimbs reduced, with ex¬
tremely short and stout radius and ulna, both pro¬
vided with large, convex distal ends. Ilia long and
square-shaped. Pubes, ischia, and femora long and
slender.
Contributions in Science, Number 416
CRANIAL SKELETON
Skull
The skull of Carnotaurus sastrei (Figs. 2-5) was
somewhat laterally compressed in fossilization,
mainly around the temporal and orbital openings.
It is short and high, with two robust frontal horns.
Both the skull and lower jaws have loose sutures
between some bones that suggest kinesis. Most of
the lateral and dorsal surfaces of the skull have
sharp rugosities, canals, and small foramina. But
the occipital region, the quadratojugal, the supra-
occipital region, and the occipital crest are smooth.
These areas of the skull were covered by a muscular
mass, whereas in the rugose areas the soft covering
possibly was horny.
The premaxilla is thick and massive, with the
narial opening in a high position. The posterodorsal
process is not exposed. Below the external nares
there is a wide depression of smooth bone, very
different from the rest of the muzzle. It may have
been the position of a salt gland comparable to that
of some herbivorous dinosaurs (Osmolska, 1979).
The maxilla is very short and high, with one
maxillary fenestra that is dorsoventrally elongated,
anteroposteriorly short, and located very near the
anterior border of the preorbital opening. The latter
opening is also higher than long, quite different
from the situation in tyrannosaurids (Fig. 7) and
dromaeosaurids.
The lachrymal has only a small anterior projec¬
tion that contacts the nasal (Fig. 2). The ventral
projection is posteriorly convex, as in Abelisaurus
(Bonaparte and Novas, 1985); fig. 7.
The jugal is very short and high, with the anterior
projection dorsoventrally thick. The suture with
Bonaparte, Novas, and Coria: Carnotaurus sastreiM^
Figure 2. Carnotaurus sastrei. Skull and left lower jaw in lateral view, an, angular; ar, articular; cor, coronoid; d,
dentary; fr, frontal; fsp, posterior surangular foramen; itf, infratemporal fenestra; j, jugal; 1, lachrymal; mx, maxilla;
mxf, maxillary fenestra; n, nasal; o, orbital opening; p, parietal; pmx, premaxilla; po, postorbital; ppo, paraoccipital
process; pra, prearticular; q, quadrate; qj, quadra to jugal; rt, retroarticular process; sa, surangular; soc, supraoccipital;
sq, squamosal.
the maxilla is rigid, whereas those with the lach¬
rymal, postorbital, and quadratojugal are weak,
suggesting that kinesis was possible.
The postorhital is proportionally large, with a
conspicuous anteroventral process that defines the
lower rim of the orbit and closely approaches the
lachrymal. The postorbital has a wide contact with
the frontal horn. The sutures with the squamosal
and the dorsal process of the jugal are weak, sug¬
gesting that movement was possible.
The squamosal is relatively small, with little par¬
ticipation of the supratemporal fenestra. The ven¬
tral projection is rod-like, with a short contact with
the quadratojugal. The cavity for the quadrate head
is not deep, and the posteroventral projection is
modest.
The quadrate and quadratojugal are fused to one
another and show a movable contact with the jugal.
The quadrate is dorsoventrally long, with the lower
condyles well defined in medial view and less so in
4 ■ Contributions in Science, Number 416
lateral view. The anteromedial projection is high,
and it is most easily distinguished at the contact
with the quadrate branch of the pterygoid. In oc¬
cipital view (Fig. 4) the quadrates are dorsoventrally
long and wide, showing a rather sharp border run¬
ning down from near the paraoccipital process, sep¬
arating the posterolateral surface from the postero¬
medial one.
The nasals show more pronounced rugosities
than do other parts of the skull. They are trans¬
versely convex and almost straight anteroposte-
riorly. The anteroventral process is pronounced.
The sutures of the nasals with the frontals, the
premaxillae, and lachrymals are weak, suggesting
mobility. But the union of the nasals with the max¬
illae, although not fused, suggests that it was more
rigid than those cited above.
The frontals are fused with the anterior part of
the parietals, but the sutures with the lachrymals
and postorbitals are loose. The horns of this species
Bonaparte, Novas, and Coria: Carnotaurus sastrei
Figure 3. Carnotaurus sastrei. Skull (A) in dorsal view with lateral compression in premaxilla, maxilla, and quadra-
tojugal, and displacement of the premaxilla corrected. Left lower jaw (B) in medial view. Abbreviations as in Figure 2,
and: e, splenial; gl, glenoid; mf, Meckelian foramen; oc, occipital condyle; pgf, post-glenoid fossa; rt, retroarticular
process; stf, supratemporal fenestra.
are formed by the frontals. They project latero-
dorsad and are relatively short and very thick, with
the dorsoposterior surfaces rather flat (Figs. 3, 4,
5B). They show a system of shallow grooves, lat-
eromedially directed, that make the surface very
different from the rugose nature of the muzzle. The
structure of the horns is not very different in ap¬
pearance from the horn cores present in bovids,
although they are certainly not identical to the lat¬
ter. The horns may have had a corneous covering,
which would have made the horns much longer in
life.
The parietals are fused to the frontals. Ante¬
riorly, they are wide, forming a flat dorsal region.
Centrally, the sagittal crest is thin and projects
abruptly dorsad to form a posterodorsal process.
This is axially thick and transversely wide, with a
posterior projection extending over the conspicu¬
ous medial crest of the supraoccipital. From here,
the supraoccipital crest extends ventrad to the squa-
Contributions in Science, Number 416
mosals. As a result, the supratemporal fenestra have
low, lateral borders, whereas medially they have
elevated borders. The anterolateral side of the pa¬
rietals diverge outward, forming a strong basal
structure for the horns and at the same time forming
the posterior orbital wall.
The supraoccipital has a prominent posterior
projection.
The occipital crest is transversely broad, al¬
though not as broad as the paraoccipital processes.
The squamosals and parietals contribute to the high
occipital crest, which in posterior view makes a
continuous subcircular dome.
The paraoccipital processes are rather thin, di¬
rected partially towards the rear and a little down¬
ward, with the lateral projections modestly ex¬
panded.
The occipital condyle is robust, subspheric, with
a marked ventral neck, and with the exoccipitals
dorsolaterally directed, although no sutures are seen.
Bonaparte, Novas, and Coria; Carnotaurus sastrei M 5
soc
Figure 4. Carnotaurus sastrei. Skull in posterior view.
Abbreviations as in Figures 2 and 3, and: boc, basioccip-
ital.
Lower Jaws
Both lower jaws (Figs. 2, 3) are completely pre¬
served with all their teeth. Unfortunately some teeth
were fragmented when the jaws were being sepa¬
rated from the skull.
The lower jaw is low and elongated (Figs. 2, 3),
with the articular region and retroarticular process
low and elongated in lateral view. It has a large
mandibular fenestra and, behind it, on the internal
side, a wide abductory fossa.
Externally, the postdentary bones have a smooth
surface, whereas the dentary shows rugosities in the
lower half, below the row of nutrient foramina.
Above the foramina the bone surface is rather
smooth.
The connection between the dentary-splenial and
the postdentary bones is extremely weak. It is re¬
duced to only two contact points, a dorsal one
between the dentary and surangular, and a ventral
one between a rod-like projection of the dentary
and the angular (Fig. 2).
The dentary bears 13 teeth and 2 empty alveoli.
The area of the jaw occupied by teeth is rather
long. The posterior area of the dentary is bifurcate
bordering a large mandibular fenestra. The pos-
terodorsal projection of the dentary has a notch
for an anterodorsal projection of the surangular.
The splenial is very near its original position
against the right dentary (Fig. 3) and a bit moved
relative to the left dentary. It is elongated and low,
with the lower border progressively thicker towards
6 B Contributions in Science, Number 416
the rear, where it forms most of the lower border
of the jaw. The dorsal border is thin and fits into
a special depression of the dentary. The anterior
extension of this bone, which is dorsoventrally short,
has an indentation that is related to the Meckelian
cartilage. The posterior margin of the splenial is
concave. The Meckelian foramen is present near
the ventral border of this bone.
The angular is relatively small, forming the ven¬
tral border of the mandibular fenestra.
The surangular borders the mandibular fenestra
dorsally and posteriorly. Its dorsal border is convex
in lateral view. Posteriorly, near the angular, it has
a 6-mm foramen, the “posterior surangular fora¬
men” indicated for Allosaurus (Madsen, 1976: pi.
8). The posterior area of the surangular is low and
elongated.
The prearticular is large, with the characteristic
open “U” shape (Fig. 3B) defining an opening be¬
tween its anterodorsal projection and the splenial.
Its posterior projection, which is rather low, reaches
the end of the retroarticular process.
The articular bears a well-developed medial pro¬
jection. The glenoid has two concave facets for the
quadrate condyles, with the axis directed antero-
mediad. The internal portion of the glenoid is larger
and more defined than the external portion, and a
clear longitudinal crest separates them. Posterior to
the glenoid facet the dorsal portion of the retroar¬
ticular process is transversely concave. There is a
postglenoid fossa between the glenoid and the dor¬
sal area of the retroarticular process (Fig. 3B).
The coronoid is a small, elongate bone placed
against the medial side of the surangular above the
prearticular. Its dorsal border is transversely wide,
and it has a lateral depression where the postero-
dorsal projection of the dentary inserts (Figs. 2,3B).
Hyoid Apparatus
Three pieces of the hyoid arch (Fig. 6) were found
articulated between and within the lower jaws.
Anteriorly there is a corpus (see Romer and Par¬
sons, 1978), which is flat, trapezoidal, and dorsally
concave. On its posterior area there are two artic¬
ular facets for the ceratobranchials (Ostrom, 1961).
These are elongate, dorsally concave, and laterally
convex.
Comparisons of Skull and Lower Jaws
Comparison with Tyrannosauridae. The general
plan of the skull and lower jaws of Carnotaurus
is very different from that of the Laurasian Creta¬
ceous carnosaurs. The genera Tarbosaurus (Ma¬
leev, 1955), Daspletosaurus and Albertosaurus
(Russell, 1970), and Tyrannosaurus (Osborn, 1912)
have the skull proportionally long and low, with
the infratemporal fenestrae reduced by an anterior
projection of the squamosal and quadratojugal. The
region of the preorbital vacuity in these genera is
anteroposteriorly large (Fig. 7K). In comparison, the
skull of Carnotaurus is proportionally short and
Bonaparte, Novas, and Coria: Carnotaurus sastrei
very high, with a wide infratemporal fenestra. The
preorbital vacuity is reduced anteroposteriorly, and
the skull has well-developed frontal horns.
The lower jaws of Carnotaurus have large man¬
dibular fenestrae, the retroarticulars are long and
low, and the contacts between the dentary and the
postdentary bones are limited.
These major differences between the Carnotau¬
rus skull and jaws and those of the Tyrannosauridae
suggest that two basically different adaptative models
existed within the role of great predators. In detail,
the main differences with the Tyrannosauridae are
as follows:
Contributions in Science, Number 416
a) Tyrannosaurids lack frontal horns.
b) The lachrymal of Carnotaurus does not have
the pronounced anterior projection wedged be¬
tween the maxilla and the nasal found in most of
the Tyrannosauridae.
c) The lachrymal of Carnotaurus is posteriorly
convex, not anteriorly convex as in the Tyranno¬
sauridae.
d) The postorbital of Carnotaurus has a very
short posterior projection, whereas in the Tyran¬
nosauridae it is long.
e) The postorbital of Carnotaurus has a con¬
spicuous anteroventral process that is very near the
Bonaparte, Novas, and Coria: Carnotaurus sastreiM?
Table 1. Measurements (in millimeters) of the skull of
Carnotaurus sastrei.
Skull
Length of the skull from extremity of
premaxilla to distal end of quadrate 596
Height of the skull from dorsal border of
occipital crest to ventral border of jugal 425
Height of the skull from tip of the horn to
ventral border of jugal 434
Distance from distal end of quadrate to top of
parietal crest 416
Distance between tip of the frontal horns 346
Greatest expanse of paraoccipital processes 245
Greatest expanse of parietal crest 185
Transverse diameter of occipital condyle 50
Preorbital opening height 172
Preorbital opening length 98
Orbit height 50
Orbit length 80
Greatest length of quadrate 221
Jaw
Total length of the jaw 595
Dentary length 337
Height of the jaw from dorsal border of
surangular to ventral border of angular 125
lachrymal, defining the ocular cavity and forming
a well-defined opening.
f) The jugal of Carnotaurus does not have the
typical anterior projection present in the Tyran-
nosauridae.
g) The squamosal and quadratojugal of Carno¬
taurus do not have the pronounced anterior pro¬
jection (always present in the Tyrannosauridae) that
virtually divides the infratemporal fenestra into two
parts. In contrast, the squamosal of Carnotaurus
has a rod-like ventral process that touches the an¬
terior edge of the quadrate.
h) In Carnotaurus the quadrate is fused to the
quadratojugal, and there is no indication of the
quadrate foramen.
i) The development of the frontal horns resulted
in strong modifications of the frontals, representing
derived characters not developed within the Ty¬
rannosauridae.
j) The pronounced axial crest of the supraoc-
cipital, which dorsally contacts the dorsoposterior
process of the parietals in Carnotaurus, is not pres¬
ent in the Tyrannosauridae.
k) The paraoccipital processes of Carnotaurus
are long and slender, whereas in the Tyrannosaur¬
idae they are massive and robust (Fig. 7M).
l) The lower jaw of Carnotaurus has a large
mandibular fenestra and only a weak contact be¬
tween the dentary and postdentary bones. These
are very different than those present in the Tyran¬
nosauridae.
m) The posterior region of the Carnotaurus
8 ■ Contributions in Science, Number 416
A B
Figure 6. Carnotaurus sastrei. Hyoid arch in dorsal (A)
and lateral (B) views, c, corpus; cb, ceratobranchial.
lower jaw is low and elongated, with the retroar-
ticular process well defined and slender. These char¬
acters are very different from those of the Tyran¬
nosauridae, where the lower jaw is high and the
retroarticular process short and heavy.
Comparison with Deinonychosauria. The ana¬
tomical differences between Carnotaurus and Dei¬
nonychosauria skulls are significant and readily ap¬
parent; however, we will discuss some of them in
detail as Carnotaurus and deinonychosaurs were
contemporaneous theropods.
The infraorder Deinonychosauria is composed
of the families Dromaeosauridae (Colbert and Rus¬
sell, 1969), with the genera Dromaeosaurus, Dei-
nonychus (Ostrum, 1969), and Velociraptor; and
Troodontidae (Currie, 1987), with the genera Sau-
rornithoides and Troodon. Both families, in spite
of good anatomical differences that support their
systematic validity (Osmolska, 1982), show a num¬
ber of common characters.
A comparison of Carnotaurus with the Dro¬
maeosauridae and Troodontidae shows outstand¬
ing differences in the skull and jaw. The more strik¬
ing differences are as follows:
a) The deinonychosaurian skull is low and elon¬
gate, very different from that of Carnotaurus.
b) The parietal crest of deinonychosaurs is rather
generalized, with plesiomorphic characters in its
Bonaparte, Novas, and Coria: Carnotaurus sastrei
Figure 7. Skull and lower jaw of Carnotaurus sastrei (A-C) compared with Abelisaurus comahuensis (D, E), Cera-
tosaurus nasicornis (F, G), Noasaurus leali (H-J), and Tyrannosaurus rex (K-M), in lateral (A, D, F, H-K), dorsal
{B, E, G, L), and posterior (C, M) views. H-J represent maxilla, squamosal, and quadrate of Noasaurus leali in lateral
view. Scale = 10 cm, except for Noasaurus leali where scale = 2.5 cm.
thickness in linear projection. In Carnotaurus the
parietal crest shows derived characters in the dorsal
projection of its posterior area, reduction of its axial
length, and anterolateral expansion as a result of
the frontal horns.
Contributions in Science, Number 416
c) The frontals of deinonychosaurs do not have
any indication of osseous crests or horns.
d) In deinonychosaurs, the posterodorsal pro¬
cess of the premaxilla contacts the nasal, the an¬
terior projection of the jugal borders the preorbital
Bonaparte, Novas, and Coria: Carnotaurus sastreiM9
depression, and the anterior projection of the lach¬
rymal has a long contact with the nasal and maxilla.
None of these characters are present in Carnotau-
rus.
e) The condition of the quadrate and quadra-
tojugal is plesiomorphic in deinonychosaurs as they
are not fused and retain the quadrate foramen. In
Carnotaurus such bones have the apomorphic con¬
dition of being fused, and the quadrate foramen is
concealed.
There are more differences between deinony¬
chosaurs and Carnotaurus. We consider those list¬
ed above sufficient to demonstrate that the skull of
Carnotaurus is structurally and morphologically
very different from that of the Deinonychosauria.
Comparison with Ceratosauridae. The compar¬
ison with Ceratosaurus (Gilmore, 1920), from the
Upper Jurassic of North America, the only known
genus of the family Ceratosauridae, results in an
interesting picture of similarities. However, at least
in reference to the skull, we do not interpret such
similarities as indicators of an ancestor-descendant
relationship, but rather as suggesting some degree
of phylogenetic relationship. The more significant
similarities are as follows:
a) The Carnotaurus premaxilla has a deep sub-
narial body, and the posterodorsal projection is
reduced or absent, as in Ceratosaurus.
b) The additional preorbital opening in Cera¬
tosaurus (according to Gilmore, 1920: pi. 18) is
small and lies quite near the preorbital opening, as
in Carnotaurus.
c) The infratemporal fenestra and the quadrate
of Ceratosaurus and Carnotaurus are dorsoven-
trally large and do not show the typical anterior
projections of the quadratojugal and squamosal of
other theropods.
d) In Gilmore (1920: fig. 53), a dorsal projection
of the parietals is seen that resembles the more
derived condition of Carnotaurus.
These similarities are accompanied by significant
differences, such as the proportions in the length
and height of the skull of each genus, the absence
in Carnotaurus of any nasal crest, and the absence
in Ceratosaurus of any indication of a frontal or
prefrontal osseous prominence. Also, in Cerato-
sauruSy the quadratojugal is not fused to the quad¬
rate, and there is a well-developed quadrate fora¬
men, whereas the opposite is true for Carnotaurus.
The contact between the mandibular bones of the
dentary and postdentary areas is rather firm in Cer-
atosauruSy but it is extremely weak in CarnotauruSy
in which the mandibular fenestra is larger than the
former.
Comparison with Allosauridae. There are a few
similarities with Allosaurus (Madsen, 1976), the best
known genus of the family Allosauridae. The most
significant of these are as follows:
a) The deep premaxilla below the external nares.
b) The dorsal development of the posterior re¬
gion of the parietals.
The differences are better expressed than the sim¬
10 ■ Contributions in Science, Number 416
ilarities. One very significant difference is that the
squamosal, paraoccipital process, quadrate, and
quadratojugal all show several derived characters
not present in Carnotaurus. The squamosal of Al¬
losaurus projects downward, the lateral end of the
paraoccipital process is lower, and the quadrate is
relatively shorter. Such morphological differences
suggest that they represent very different adaptative
types, although they may share a common ancestor
older than Late Jurassic.
In the lower jaw, the differences between Car¬
notaurus and Allosaurus are very prominent be¬
cause the mandibular fenestra is almost absent in
the latter genus (Madsen, 1976: pi. 1).
Comparison with Abelisauridae. Abelisaurus
comahuensis Bonaparte and Novas, 1985 (Fig. 7D,
E), the type genus of the family Abelisauridae, from
the Allen Formation, Lower Maastrichtian of Rio
Negro Province, Argentina, is a rather large thero-
pod, with a skull of some 85 cm in total length.
Although the skull of Carnotaurus has different
proportions than the skull of Abelisaurus (short
and high in the former, low and elongated in the
latter), there are several common features that sup¬
port allocation to the same family. They are the
following:
a) The squamosal is very similar, especially in
the shape and orientation of the ventral process,
but with some differences in the posterior process.
b) The quadrate in both genera is long and fused
to the quadratojugal.
c) The postorbital almost meets the lachrymal
underneath the ocular cavity, although there are
morphological differences in the anterior process
of this bone between the genera.
d) The lachrymal is dorsoventrally convex to¬
wards the rear in each genus, and it contacts the
postorbital above the orbit.
e) The shape and location of the additional
preorbital opening is the same in both genera.
f) The premaxilla is deep below the narial open¬
ing in both genera.
g) The nasals show rugosities along most of the
dorsal and lateral sides in both genera.
h) The posterior region of the parietal crest is
elevated, and the supraoccipital bears a large dor-
soventral keel in the axial plane, in both genera.
i) In dorsal view, the shape of the parietal crest
is narrow posteriorly, widening anteriorly in both
genera.
Of the common characters listed above, we con¬
sider that b, c, d, e, g, and i are synapomorphies of
the family Abelisauridae. Abelisaurus and Carno¬
taurus are thus placed together in this family,
AXIAL SKELETON
Vertebral Column
The vertebral column of specimen MACN-CH 894
is complete and articulated from the atlas through
the 6th caudal. The posterior caudals were de-
Bonaparte, Novas, and Coria: Carnotaurus sastrei
Figure 8. Carnotaurus sastrei. Atlas and axis in lateral view (A) and axis in dorsal view (B). ati, atlantal intercentrum;
axi, axial intercentrum; dp, diapophysis; ep, epipophysis; ne, neurapophysis; ns, neural spine; od, odontoid process; pi,
pleurocoel; pnc, pneumatic cavity; poz, postzygapophysis; pp, parapophysis; prz, prezygapophysis. Hatched area rep¬
resents break; dashed lines represent estimated reconstruction.
stroyed by weathering, except the 12th, which is
incomplete.
In general terms, the sequence of vertebrae is
characterized by opisthocoelous cervicals with very
reduced neural spines and very large epipophyses,
and most of the dorsals are subamphiplatyan. The
sacrum has seven fused vertebrae, and the anterior
caudals have transverse processes projecting dor-
solaterally.
Cervical Vertebrae. The atlas (Fig. 8A) has the
intercentrum almost complete, and it is fused to
the atlantal arches. The articular facet for the oc¬
cipital condyle spreads onto most of the dorsal
articular area and projects upward and forward. It
does not show the articular facet for the odontoid
as is seen in Deinonychus (Ostrom, 1969: fig. 26)
but resembles the characters of Ceratosaurus (Gil¬
more, 1920: pi. 19) and Allosaurus (Madsen, 1976:
pi. 11). In the ventral side of the intercentrum there
are two processes (one on each side) for muscular
attachments. On the posterior face the articular area
for the axis intercentrum, which has a good suture
with the centrum of the axis, is transversely wide,
resembling the condition in Ceratosaurus and to a
lesser extent that in Allosaurus. The atlantal arch
is fused to the intercentrum and shows no indica¬
tion of articulation with a proatlas, which may not
have existed in Carnotaurus. The posterior pro¬
jection is incomplete; the postzygapophysis is small.
In posterior view the lower side of the roof made
by the atlantal arches has a subcircular cavity for
the passage of the spinal chord. Mobility between
Contributions in Science, Number 416
the atlas and axis appears to have been small in this
genus.
The axis (Fig. 8A, B) is complete and shows little
deformation. The centrum and intercentrum are
relatively elongated, but with rather modest trans¬
verse and dorsoventral diameters.
In anterior view the continuous surface for ar¬
ticulation with the atlas and the odontoid articu¬
lation for the occipital condyle are seen. The pos¬
terior face of the centrum is very concave, with the
lower area projected posteriorly. In ventral view
the axis bears a poorly defined, rather vestigial keel.
In lateral view two small pleurocoels are present in
the upper half of the centrum. A well-defined cav¬
ity, subdivided into two, lies behind and a bit dorsal
to the diapophyses, and two depressions with fo¬
ramina are located above and behind the prezyg¬
apophysis. The depressions with foramina that pen¬
etrate the axis probably correspond to a system of
pneumatic cavities as in Piatnitzkysaurus floresi
(Bonaparte, 1986a).
The prezygapophysis is small, whereas the post¬
zygapophysis is large with the longest axis trans¬
versely placed. The parapophysis is not defined,
although it may correspond to the suture between
the intercentrum and the centrum. The diapophysis
is rather small.
The blade of the neural spine is large with the
dorsal border convex in lateral view, as in Cera¬
tosaurus (Gilmore, 1920: pi. 19). This border is
slightly forked at the posterdorsal end in dorsal
view. Below it a division follows to the epipophy-
Bonaparte, Novas, and Coria: Carnotaurus sastreiM 11
Figure 9. Carnotaurus sastrei. Third cervical vertebra in dorsal (A), lateral (B), and anterior (C) views. Outline of
neural spine in lateral view indicated by dashed lines. Abbreviations as in Figure 8, and: cn, neural canal; dn, depression.
seal area. In posterior view there is a deep depres¬
sion with foramina in the axial plane below the
neural spine and a wide, deep depression with fo¬
ramina above the neural canal. The axis is markedly
different in almost all details from that of Tyran¬
nosaurus (Osborn, 1917). It is, however, very sim¬
ilar to that of Ceratosaurus and, to a lesser degree,
Allosaurus (Madsen, 1976: pi. 11).
The third cervical (Fig. 9) is complete and, like
the remaining cervicals, is characterized by a sig¬
nificant reduction of the neural spine and hyper¬
trophy of the epipophysis.
In lateral view the anterior face of the centrum
makes a sharp angle with the anteroposterior plane,
indicating the limits of the dorsoventral flexion of
the neck.
There are two small pleurocoels on the rather
flat lateral side of the centrum. The lower border
of the centrum is almost straight. Behind and above
the diapophysis there is a large depression and a
foramen passing into the neural arch. Anterior to
the diapophysis there is a modest foramen. Both
12 ■ Contributions in Science, Number 416
the parapophysis and diapophysis are of modest
size. The epipophysis is strongly developed, and a
lamina connects it to the prezygapophysis. This
lamina sharply delimits the lateral and dorsal areas
of the neural arch. In lateral view, the neural spine
is a bit lower than the epipophysis. The prezyg-
apophyses are well separated from one another,
each having the major axis transversely directed and
each inclined towards the axial plane.
In posterior view a large cavity is exposed that
leads into the neural arch. In the axial area there
is a rugose zone for insertion of intervertebral lig¬
aments. The posterior concavity of the centrum is
deep, with its geometric center moved to the upper
half. In dorsal view this vertebra is rather flat with
wide laminar surfaces.
The fourth cervical is basically the same as the
3rd cervical, but larger. The ventral border of the
centrum is more curved, the parapophysis better
defined, and the diapophysis thicker. The cavities
and foramina for the pneumatic system are similar,
except that the one behind the diapophysis is larger.
Bonaparte, Novas, and Coria: Carnotaurus sastrei
Figure 10. Carnotaurus sastrei. Sixth cervical vertebra in dorsal (A), lateral (B), and anterior (C) views. Abbreviations
as in Figures 8 and 9, and: pnf, pneumatic foramen.
The postzygapophyses are larger and the epipo-
physes are axially longer, with a posteriorly well-
defined process and an incipient anterior process.
In anterior view the neural spine is more reduced
than in the 3rd cervical. There is a depression on
the dorsal surface in front of the reduced neural
spine and another one between the neural spine
and the dorsal projection of the epipophysis.
The fifth cervical is larger than the 4th cervical,
especially in the volume of the centrum. The di¬
apophysis is more robust, and the epipophysis has
the anterior and posterior processes more defined.
Also, the lamina that extends between the epipo¬
physis and prezygapophysis is higher and more lat¬
erally placed. The system of depressions and fo¬
ramina of the pneumatic cavities is similar to that
of the 4th cervical.
Contributions in Science, Number 416
The sixth cervical (Figs. 10, 11) is a little larger
than the 5th cervical. In the centrum the inclina¬
tions of the anterior and posterior faces are more
pronounced. The epipophyses are better developed
anteroposteriorly, and the postzygapophyses are well
separated.
The seventh cervical is essentially the same as
the 6th cervical, although more robust. The post-
zygapophysis is larger, with the larger dimension
transversely placed and the articular facet directed
obliquely ventromediad. In posterior view the wide
depression behind the neural spine has two rather
large fenestrae that lead into the neural arch. Pos¬
teriorly, the centrum is subcircular and less concave
than in the 6th cervical.
The eighth cervical vertebra is more voluminous
and higher than the 7th cervical. The centrum is
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 13
Figure 11. Carnotaurus sastrei. Sixth cervical vertebra
in posterior view. Scale = 10 cm.
slightly concave on its ventral side, as in the pre¬
ceding ones, and it lacks a keel. In the area of the
pleurocoel there is a depression and three foramina
penetrate the centrum. The neural spine is almost
nonexistent, and the epipophysis is more elevated
than in the 7th cervical. The diapophysis is a little
higher, and the opisthocoelia is less pronounced.
The ninth cervical (Fig. 12) is larger than the 8th
cervical, although it shows generally the same mor¬
phology. In comparison, however, the opisthocoe¬
lia is less pronounced, the diapophyses are more
elevated and stronger, and the neural spine is a little
more developed. The system of depressions, foram¬
ina, and pneumatic cavities appears to be the same,
but with larger foramina. A ventral, rounded keel
is present on the centrum, with slight depressions
on each side of it.
The tenth cervical (Fig. 13) has a centrum with
the anterior and posterior articulations at a right
angle with respect to the horizontal axis. A modest
keel exists on the ventral border. The diapophysis
is more elevated, more robust, and longer than in
the 9th cervical. The neural spine is also higher,
although it remains lower than the epipophysis. The
latter shows some reduction, with the anterior and
posterior processes very reduced. On the posterior
side of the neural spine there are strong rugosities
for the attachment of interspinous ligaments.
Dorsal Vertebrae. The first dorsal vertebra was
affected by a small fault in the rock that produced
some displacement and deformation in its neural
arch. The centrum is almost the same, except for
a larger “pleurocoelus depression,” and the par-
apophysis is more dorsally placed.
In the neural arch the changes in relation to the
last cervical are abrupt. There is no epipophysis,
the neural spine is the highest dorsal element of the
vertebra, and the robust transverse process is at the
level of the zygapophyses. However, the morphol¬
ogy and orientation of the zygapophyses are similar
to those of the cervicals. The total height of the
vertebra is greater than the 10th cervical. Compar¬
atively, the shallow opisthocoelia of the last cervical
and first dorsal are very different to the marked
opisthocoelia of the same vertebrae of Allosaurus
(Madsen, 1976: pi. 15).
The second dorsal (Fig. 14) has the transverse
process more elevated and robust than in the 1st
dorsal, and the neural spine is a little larger. The
zygapophyses are nearer to the axial plane, in par¬
ticular the postzygapophyses, which show an incip¬
ient hyposphene. In lateral view, two borders are
present on the ventral side of the transverse process:
the anterior and posterior infradiapophysial lami¬
nae. A noticeable depression penetrating into the
neural arch is present between the anterior infra¬
diapophysial lamina and the prezygapophysis.
The third dorsal is like the 2nd dorsal, but the
neural spine is longer and more robust and the
transverse process longer. The zygapophyses are
less inclined, particularly the postzygapophysis,
which is almost horizontal. Below the postzyg¬
apophysis is a well-developed hyposphene. Under¬
neath the prezygapophysis there is a column-like
process, which borders laterally a large depression
that communicates with the neural arch.
The fourth dorsal (Fig. 15) has a longer neural
spine than the 3rd dorsal, which is also anteropos-
teriorly flat and posteriorly inclined. The prezyga¬
pophysis is subhorizontal and projects anteriad. The
postzygapophysis is horizontal, and the lateral bor¬
der is directed downward. The hyposphene of this
vertebra is well developed and forms a wide body
that reaches the roof of the neural canal. The par-
apophysis is dorsal to the vertebral body.
In the fifth dorsal (Fig. 16) the parapophysis is in
a very different position—-approximately in the
middle of the neural arch and projecting laterad.
The parapophysis borders a large cavity that pen¬
etrates into the neural arch, in contrast with the
condition in the 4th dorsal. The neural spine is
large, in part laterally compressed, and with a dorsal
expansion. The centrum is almost amphiplatyan.
In the sixth dorsal, and to a degree in the 5th
dorsal, below the prezygapophysis there is a lower,
anterior projection. The parapophysis is more dor-
sally placed, and it has a long, laterally projecting
stem. Between the parapophysis stem and the di¬
apophysis there is a very large fossa communicating
into the neural arch. In dorsal view the transverse
process has a wide, flat surface. The centrum of
this vertebra has a lateral depression oriented
anteroposteriorly just below the union of the cen¬
trum with the neural arch. Two conspicuous fo¬
ramina are located there. Along most of its perim¬
eter the posterior border of the centrum is
pathologically hypertrophied.
The seventh, eighth, and ninth dorsals (Figs. 17,
18) are of similar morphology. The 7th dorsal is a
bit taller than the 6th dorsal, and the process below
the prezygapophysis persists. The postzygapophysis
is transversely concave. In the neural spine, there
are strong osseous processes for the attachment of
14 ■ Contributions in Science, Number 416
Bonaparte, Novas, and Coria: Carnotaurus sastrei
c
Figure 12. Carnotaurus sastrei. Ninth cervical vertebra in lateral (A), anterior (B), and posterior (C) views. Scale =
10 cm.
tendons. In the 8th dorsal, the dorsal surface of the
transverse processes is larger than in the 7th dorsal,
and it is even more so in the 9th.
The tenth and eleventh dorsals (Figs. 19,20) have
centra of greater diameter than the preceding ver¬
tebrae. The parapophysis, very dorsal in position,
tends to become part of the dorsal lamina of the
diapophysis. The parapophysis, is larger anteropos-
teriorly than in previous vertebrae, increasing its
surface posteriorly. The neural spine of the 10th
dorsal does not show the conspicuous osseous ru¬
gosities for attachment of tendons, but they are
present in the 11th. The prezygapophysis of the
11th dorsal is reduced and suggests that the move¬
ments between the 10th and 11th vertebrae were
very restricted.
Sacral Vertebrae. The sacrum of Carnotaurus is
almost complete, except for the system of plate¬
like sacral ribs that was not well preserved. The
Contributions in Science, Number 416
sacrum consists of seven vertebrae, of which the
1st sacral is fused only in the vicinity of the post-
zygapophysis and the following six are intimately
fused (see Fig. 20). The 1st sacral preserves most
of the characters of the last dorsals, but the trans¬
verse processes are reduced, both in length and
width. The neural spine is axially reduced, and the
lateral cavities of the neural arch are concealed.
The 2nd to 7th sacral vertebrae are strongly fused,
both through the centra and through the neural
arches. The neural spines, especially distally, form
a continuous ossification. The co-ossified centra
form, in lateral view, a dorsally convex arch. In
ventral view, the centra in the middle of the sacrum
(sacrals 3, 4, and 5) are transversely reduced, par¬
ticularly the 4th sacral. There is thus a marked
reduction of volume in the vertebral bodies of the
sacrum.
The lateral projections of the sacrum are short.
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 15
Figure 13. Carnotaurus sastrei. Tenth cervical vertebra in dorsal (A), lateral (B), and anterior (C) views. Abbreviations
as in Figures 8 and 9.
indicating that the pelvis of Carnotaurus was nar¬
row. The main connection between the sacrum and
the ilia was through the transverse processes. The
sacral ribs are incomplete, but appear to be very
short. From the 3rd sacral distad they are rather
thick in cross section.
Caudal Vertebrae (Figs. 21-23). Only six proxi¬
mal, articulated caudals are preserved, with some
of them showing effects of weathering. Also pre¬
served is an isolated centrum with part of the neural
arch, possibly the 12th caudal.
The first two caudals are very large and robust.
The caudals rapidly decrease in size posteriorly to
the 5th, and the centrum of each is amphicoelous.
Elongation of the vertebral bodies is apparent, and
the centra show only slight lateral depressions.
The dorsolaterally projecting transverse process¬
es reach a level very near the top of the neural
spine, suggesting a derived condition. The lower
surface of the transverse process faces ventrolat-
erally. Another curious, derived character seen in
the available caudals is the expanded distal end of
the transverse process. This feature is present in the
16 ■ Contributions in Science, Number 416
six proximal caudals, whereas in the 12th caudal it
is not preserved.
Comparison of Vertebral Column. The presacral
vertebrae of Carnotaurus sastrei correspond well
with the general model present in the Theropoda,
particularly in the Carnosauria. However, they bear
several derived characters that readily distinguish
them from the known theropods, except perhaps
Noasaurus kali (Bonaparte and Powell, 1980).
The cervicals of Carnotaurus, posterior to the
axis and up to the 10th, show a strong reduction
of the neural spine and a strong dorsal development
of the epipophyses, which form a paired row of
neural spine-like processes. This model of cervical
vertebrae was undoubtedly adapted for quite a new
arrangement of the muscular system, related to bet¬
ter functional control of neck and cranial move¬
ments. It resembles in part the neck morphology
of the Ornithomimidae, in which the neural spine
is vestigial and bears dorsolateral processes higher
than the neural spine.
In Allosaurus fragilis the 4th and 7th cervicals
show the lateral lamina connecting the epipophysis
Bonaparte, Novas, and Coria: Carnotaurus sastrei
with the prezygapophysis, and the 5th cervical shows
good dorsal development of the epipophysis (Mad¬
sen, 1976: pL 13), although it is less developed than
in Carnotaurus, However, the anatomy of these
vertebrae of Allosaurus suggests that, at least in
part, the unique anatomy of the cervicals of Car¬
notaurus were roughly outlined in Allosaurus.
The dorsal vertebrae are significantly different in
the lateral structure of the neural arch from those
of the Tyrannosauridae, Allosauridae, Ceratosaur-
idae, and Deinonychus. The differences relate to
the higher position of the parapophyses in Car¬
notaurus, which increases from the 5th dorsal pos¬
teriorly to reach their highest point in the 10th and
11th dorsals, in which diapophyses and parapoph¬
yses are very near one another in the horizontal
plane. Thus, the dorsal migration of the parapoph¬
yses of Carnotaurus led to several autopomorphies
Contributions in Science, Number 416
of the Patagonian form, i.e., anteroposteriorly wide
transverse processes, presence of a discrete fossa
between the parapophyses and the infradiap-
ophysial lamina, and presence of a well-defined
lamina between the parapophysis and dorsolateral
border of the centrum.
In contrast to the similarities of the skull and axis
discussed previously, no shared derived characters
have been seen in the presacral vertebrae of Car¬
notaurus and Ceratosaurus.
The count of seven sacral vertebrae and the pres¬
ence of a six-vertebrae synsacrum represent signif¬
icant derived characters not recorded, as far as we
know, in other Carnosauria. However, a compa¬
rable number of sacrals is present in some small
theropods, such as Saurornithoides (Barsbold,
1974). Osborn (1917) recognized five sacral verte¬
brae in Tyrannosaurus, although from the figure
Bonaparte, Novas, and Coria: Carnotaurus sastreiM 17
Figure 15. Carnotaurus sastrei. Fourth dorsal vertebra in dorsal (A), lateral (B), and anterior (C) views. Abbreviations
as in Figures 8 and 9, and: hy, hyposphene.
of the pelvis and sacrum in lateral view we suspect
that the 13th dorsal may very well be the first func¬
tional sacral because of its position “within” the
ilia, even though it is not fused to the remaining
sacrals. In Osborn’s paper (1917: fig. 19) one can
see that the sacrum of Tyrannosaurus is propor¬
tionally shorter than that of Carnotaurus, and it is
wider at both ends. Also, it is possible to see that
it has neither the dorsal arching along the ventral
border of the centra nor along the row of sacral
ribs. Even though the co-ossified dorsal end of the
neural spines is a common character for Tyran¬
nosaurus and Carnotaurus, the arching of the cen¬
tra and the line of union between the sacrum and
ilia (see Fig. 20), as well as the strong co-ossification
of the centra and their reduced thickness, represent
a set of derived characters not present in Tyran¬
ts ■ Contributions in Science, Number 416
nosaurus. This suggests that the sacrum of the Pat¬
agonian genus is more derived than that of the
tyrannosaurids.
Piatnitskysaurus and Allosaurus clearly have
more primitive sacra (Gilmore, 1920; Madsen, 1976;
Bonaparte, 1986a), each with five vertebrae that are
not completely fused.
Ceratosaurus shows similarities in the general
plan and in several derived characters. The sacrum
is composed of five fused vertebrae (Gilmore, 1920),
which are well co~ossified, as in Carnotaurus. In
addition, in Ceratosaurus reduction of the thick¬
ness of the fused centra is also similar to the situ¬
ation in Carnotaurus. Gilmore considered the sa¬
cral vertebrae of Ceratosaurus to number five, taking
into account only the fused ones. In our opinion
the functional sacral vertebrae of this North Amer-
Bonaparte, Novas, and Coria: Carnotaurus sastrei
c
Figure 16. Carnotaurus sastrei. Fifth dorsal vertebra in anterior (A), posterior (B), and lateral (C) views. Scale =
10 cm.
ican Morrison Formation genus includes seven and
possibly eight vertebrae, including the last two pre-
sacrals and possibly the hrst caudal of Gilmore. The
seven sacral vertebrae of CarnotauruSy with the 1st
sacral “recently” incorporated from the dorsals (note
the intermediate morphology of this vertebra and
its partial fusion with the posterior sacrals [see Fig.
20]), suggest that general organization of the sacrum
is rather similar to that of Ceratosaurus. The sim¬
ilarities in the sacrum of these two genera strongly
suggest a common trend of specialization that may
reflect a close phylogenetic relationship.
Concerning the caudal vertebrae, we are not aware
of comparable characters in other theropods. It
suggests that both uncinate processes and laterally
elevated transverse processes may represent an aut-
apomorphy of Carnotaurus.
Contributions in Science, Number 416
Ribs
The complete series of cervical ribs and most of
the dorsal ribs are preserved. The atlantal rib (Fig.
24A) has an articular head with a slightly flat ex¬
pansion, but without indication of a separate ca-
pitulum or tuberculum. The stem is long, rather
flat, and reaches the 4th cervical vertebra.
The second rib is larger and has a more defined
proximal area. The shaft is rather flat until the 3rd
vertebra, behind which it extends stiliform to the
5th cervical vertebra.
The third cervical rib (Fig. 24B) shows a well-
defined capitulum and tuberculum. The latter forms
an expanded dorsal process. A short anterior pro¬
cess that increases in size in the following ribs is
present. The shaft of this rib is very long, almost
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 19
Figure 17. Carnotaurus sastrei. Eighth dorsal vertebra in dorsal (A), lateral (B), and anterior (C) views. Abbreviations
as in Figures 8, 9, and 15, and: hp, hypanthrum; li, scar for interspinous ligaments.
reaching the 7th cervical vertebra. The proximal
area of the shaft is laterally convex and continues
distad as a subcylindrical rod of constant diameter.
The fourth cervical rib (Fig. 24C) is similar to
the previous one, but the proximal laminar area is
larger, followed posteriorly by an elongate and del¬
icate stilet.
The shape of the 4th cervical rib is repeated up
to the 9th cervical rib (see Fig. 24D, E). The anterior
20 ■ Contributions in Science, Number 416
process remains conspicuous, and each rib pos¬
sesses a well-defined capitulum and tuberculum.
In the tenth cervical rib a lamina connects the
capitulum and tuberculum. The anterior process is
abruptly reduced, and there is no distinction in the
shaft between the proximal and distal portions. The
shaft of this rib is relatively short and thick.
The first dorsal rib (Fig. 25A, B) is much longer
and thicker, and it has a large capitulum and robust
Bonaparte, Novas, and Coria: Carnotaurus sastrei
Figure 18. Carnotaurus sastrei. Eighth dorsal vertebra
in posterior view. Scale = 10 cm.
tuberculum. The second dorsal is similar, but larg¬
er, with a larger and stronger proximal process. The
third and fourth dorsal ribs (Fig. 25C, D and E, F,
respectively) increase in size. The size of the prox¬
imal region of each rib decreases posteriorly, and
the shaft of the capitulum becomes shorter because
of the dorsal migration of the parapophysis. The
last dorsal, the 11th (Fig. 251, J), is short and in the
living animal would have almost touched the an¬
terior border of the ilia.
Gastralia
Because of weathering and the nature of the ex¬
cavation process, the gastralia, which were in their
original position, were not properly recovered. They
were positioned from very near the anterior border
of the pubis forward and consisted of rather long,
subcylindrical rods, lying transversely very near one
another.
Haemal Arches
The anterior haemal arches (Fig. 26) are represented
by natural molds and bone fragments (one with the
proximal part preserved and another one complete).
They are proportionately long and slender, with a
relatively wide proximal articulation and the hae¬
mal canal enclosed dorsally. The articular facets
and the shape of the haemal canal suggest that the
haemapophyses had a more posterior than ventral
orientation.
In the anterior part, the haemal canal has a lateral
expansion, whereas posteriorly the canal shows a
deep distal depression. These features indicate that
the haemal canal had an oblique position relative
to the long axis of the shaft of the haemapophysis.
These features suggest that natural position of the
haemapophysis was far from perpendicular to the
Table 2. Measurements (in millimeters) of presacral, sa¬
cral, and caudal vertebrae of Carnotaurus sastrei
Vertebra
number
Maximum
length
of centra
Maxi¬
mum
width
(ante¬
rior) of
centra
Maximum
height
(anterior)
of centra
Greatest
height
overall
Presacrals
Atlas
45
83
40
95a
Axis
118
84
55
198
3
100
68
41
175t
4
no
71
50
187t
5
119
82
56
210at
6
120
92
67
225t
7
no
97
70
226t
8
108
105
87
250t
9
104
111
96
265t
10
98
115
106
245t
11
100
114
—
256
12
101
112
108
257
13
103
120
—
—
14
108
no
105a
276
15
101
101
101
293
16
117
106
106
315a
17
123
120
105
330
18
122
117
117
343
19
120
127
115
360
20
116
137
120
366
21
120
145
120
365
Sacrals
1
132
142
—
382
2
115
130
—
346
3
112
70
—
305
4
98a
52
—
288
5
71a
42
—
302
6
118
56
—
324
7
124
82
—
358
Caudals
1
128a
140
126a
346
2
122
144
118
341
3
120
—
120
326
4
136
104
117
324
a = approximate.
t = from dorsal border of epipophysis to posteroventral
border of the centrum.
axis of the tail and that it more probably formed
a very acute angle with it. It also means that the
tail of Carnotaurus was probably dorsoventraly
flattened.
APPENDICULAR SKELETON
Pectoral Girdle and Forearm
Scapulo-Coracoid and Sternal Plates. Both scap-
ulo-coracoids are completely preserved. The cor-
Contributions in Science, Number 416
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 21
Figure 19. Carnotaurus sastrei. Tenth dorsal vertebra in dorsal (A), lateral (B), and anterior (C) views. Abbreviations
as in previous figures.
acoid and the proximal part of the scapula are rath¬
er large compared with the modest scapular blade
(Fig. 27). The latter has a constant width from a
little above the glenoid cavity to the distal end,
where there is no expansion as in Tyrannosaurus
(Osborn, 1917) or other carnosaurs.
The anterior border of the scapula passes grad¬
ually to the acromial area, resembling the condition
of DeinonychuSy whereas it is more angular in Ty¬
rannosaurus, Allosaurus, and Ceratosaurus.
The scapula and coracoid are strongly fused. The
22 ■ Contributions in Science, Number 416
coracoid is large, with the distal border making a
wide, continuous curvature, from near the pos¬
terolateral process of the coracoid up to the scap-
ulo-coracoid suture on the anterior border. The
position of the posterolateral process is near the
glenoid cavity, a character we have not seen in other
Camosauria. In this area there are rugosities sug¬
gesting strong muscular attachment, probably of
the m. biceps.
At the level of the glenoid cavity there is a large
subacromial depression, located below a border of
Bonaparte, Novas, and Coria: Carnotaurus sastrei
Figure 20. Carnotaurus sastrei. Sacrum and both ilia in dorsal (A), ventral (B), and lateral (C) views, ab, anterior
blade; ac, acetabulum; Dll, 11th dorsal vertebra; ip, ischiatic peduncle; pb, posterior blade; pu, pubic peduncle; SI
and S7, 1st and 7th sacral vertebra.
the scapula that may be the acromial process. The
coracoid foramen is large and perforates the cor¬
acoid with a slight inclination.
The glenoid cavity is well defined, with well-
developed supra- and infraglenoid processes that
Contributions in Science, Number 416
enlarge the articular region dorsally and ventrally.
The shape is subspheric, except in the posterior
section of the coracoid where the cavity between
the processes is flat.
Near the area of the glenoid cavity in both scap-
Bonaparte, Novas, and Coria: Carnotaurus sastreiM23
A
Figure 21. Carnotaurus sastrei. First caudal vertebra in dorsal (A), lateral (B), and anterior (C) views. Abbreviations
as in previous figures, and: tr, transverse process.
ulo-coracoids there is an ossification on the inner
side of the bones (about 50 mm in diameter), the
function of which is unknown to us.
In life, the scapular blade was probably almost
parallel with the row of dorsal vertebrae. On the
internal side of the coracoid there are marks of
dorsal ribs, indicating that they were parallel to the
long axis of the coracoid.
The clavicle (Fig. 27C) may be represented by a
slightly curved, rod-like bone with a modest, prox¬
imal expansion. It was lying on the anteroexternal
24 ■ Contributions in Science, Number 416
portion of the right coracoid, in front of the sternal
plates. Because of its position and morphology we
interpret it as the right clavicle, although some por¬
tion of it may be missing.
The sternal plates (Fig. 27C) are preserved nearly
complete. They were found close behind the cor¬
acoids. The right humerus was lying on the lateral
surface of the right sternal plate. They are nearly
flat and oval in outline, with a short ventral pro¬
jection and two rib processes projecting postero-
dorsad and separated by a concavity.
Bonaparte, Novas, and Coria: Carnotaurus sastrei
The position of the sternal plates as found, i.e.,
touching the medial border of the coracoids, may
not have been too far from their natural position.
Forelimbs. The two forelimbs are almost com¬
plete, articulated, and with both humeri within the
glenoid cavity. Unfortunately, the forefeet are not
completely articulated, and some bones are not easy
to interpret.
The forelimbs of Carnotaurus have significant
derived characters, particularly in the epipodial
bones, which are strongly reduced and appear to
be a functional part of the hand. In this, the Pat¬
agonian genus shows derived characters not re¬
corded in other Theropoda.
The humerus (Fig. 28) is robust and relatively
short. Its dorsal region is transversely convex, and
the internal side is concave proximodistally, as is
common in the Carnosauria, and the opposed side
is convex in the same direction. This means that in
dorsal view the humerus is curved distally and in¬
ternally, and in lateral view relatively straight.
The humeral head is well defined and subspheric,
with similar length axes in the dorsoventral and
lateromedial direction. A well-defined, but short,
neck is present. These characters suggest that the
humeral movements were pronounced in all direc¬
tions, although the anteroposterior component was
probably greater. The internal tuberosity is well
developed, although proportionally smaller than in
Piatnitzkysaurus (Bonaparte, 1986a) or Allosaurus
(Madsen, 1976). The deltopectoral crest is massive,
with the distal area most pronounced. Distally, the
articulations for radius and ulna are dorsoventrally
large and relatively flat, suggesting that the forelimb
bones did not rotate on them significantly and that
movements were probably restricted to the antero¬
posterior plane. Several typical characters of the
theropod humerus have been strongly modified in
the distal area of the Carnotaurus humerus: a) the
large supracondyloid depression on the dorsal face
was replaced by a convexity of the bone; b) the
intercondylar depression on the ventral side was
also modified, and this area is convex; c) the artic¬
ular condyles were modified from subspheric to flat
and dorsoventrally large; and d) the proximodistal
torsion of the humerus is notably reduced.
The radii and ulnae (Figs. 29A-C, 30) are com¬
plete, and in articulation with the humeri and each
other. They are very short, each only A the length
of the humerus. The ulna has a large, concave ar¬
ticular facet for the humerus, with the posterior
area elevated and transversely wide. The anterior
part of the facet is narrow and in a lower position
than the posterior. In proximal view, the articular
facet is triangular, with the apex anterior. The distal
aniculation is a wide condyle, convex in both di¬
rections, with the longest axis directed anteropos-
teriad. A marked neck lies between the distal ar¬
ticulation and the short diaphysis.
The radius is slightly shorter than the ulna. It has
a large, rather flat, proximal articular facet. It ar¬
ticulates with the ulna via a convex area near its
Contributions in Science, Number 416
Figure 22. Carnotaurus sastrei. Second caudal vertebra
in posterior view. Scale = 10 cm.
proximal end. The distal articulation is large and
wide in both directions, but less convex than that
of the ulna. There is not as great a difference be¬
tween the size of the shaft and the ends of the bone
as seen in the ulna. A well-defined osseous process,
whose function is unknown, is found on the lateral
side of the diaphysis and in the middle of its length.
Large parts of both the right and left manus have
been preserved, although we do not have a defin¬
itive interpretation of the distribution or relation¬
ship of the different pieces of their carpi and digits
of the Carnotaurus. The reconstruction is tentative
(Fig. 29C); however, we observe the following:
a) There is a group of carpal bones, of unknown
number, below the ulna.
b) Four metacarpals are present. Metacarpal I is
short, similar in length to metacarpal III; metacarpal
II is the largest, morphologically similar to I, but
Va longer in size, with a large proximal articulation;
metacarpal III is shorter than II, but transversely
wider than metacarpals I and II, with the proximal
area anteroposteriorly large and the distal area
smaller; metacarpal IV is atypical, with a very large
articulation for the large condyle of the ulna.
c) The first row of phalanges was limited to digits
I, II, and perhaps III. The second row of phalanges
is represented only by a proximal fragment of the
second phalanx of digit 11.
Comparisons of Pectoral Girdle and Fore¬
limb. A comparison of the scapulo-coracoid of
Carnotaurus with that of the genera of the families
Allosauridae, Tyrannosauridae, and Ceratosauridae
shows strong differences. The weakly developed
acromial region of Carnotaurus differs greatly from
that in the cited families, in which this region is
well developed. Also, in Carnotaurus the postero¬
lateral process of the coracoid is very near the
glenoid cavity.
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 25
Figure 23. Carnotaurus sastrei. Neural arch and incomplete centrum of the 6th caudal vertebra in dorsal (A), lateral
(B), and anterior (C) views. Abbreviations as in previous figures.
The scapular blade resembles more of the dro-
maeosaurid type of scapula, particularly that of Dei-
nonychus, with the parallel anterior and posterior
borders and the modest acromial expansion. How¬
ever, differences with the coracoid of Deinonychus
are strong enough to minimize any phylogenetic or
systematic interpretations based on the similarities
of the scapular blade.
The humerus of Carnotaurus is very similar to
26 ■ Contributions in Science, Number 416
that of Tyrannosaurus (Osborn, 1917: fig. 21) and
Albertosaurus (Lambe, 1917: fig. 31), although there
are clearly differences in details, such as a more
reduced deltopectoral crest, a more developed in¬
ternal tuberosity, and a more derived distal region
in the humerus of Carnotaurus.
The megalosaurid Torvosaurus (Galton and Jen¬
sen, 1979) has a different model of humerus with
respect to Carnotaurus in the development of the
Bonaparte, Novas, and Coria: Carnotaurus sastrei
deltopectoral crest and humeral head. Carnotaurus
has radii and ulnae proportionally shorter than Tor-
vosaurus.
Figure 25. Carnotaurus sastrei. Proximal ends of right
dorsal ribs. 1st rib (A, B), 2nd rib (C, D), 4th rib {E, F),
9th rib, (G, FI), 11th rib (I, J) in ventral (A, C, E, G, I)
and dorsal (B, D, F, H, J) views.
Contributions in Science, Number 416
of the several bones observed, and the characters
of the distal end of the radius and ulna, are not
typical for a carnosaur. The hand shows a group
of plesiomorphic (c.g., four digits) and apomorphic
(e.g., metacarpal IV with a very large articulation
for the distal condyle of the ulna) characters, whose
association has not been reported in other Therop-
oda, except in Ceratosaurus where Gilmore (1920:
figs. 60, 62) illustrated four digits with short pha¬
langes.
We believe that the strong modification in the
forelimbs of Carnotaurus, among them the almost
total lack of torsion of the humerus and the parallel
position of radius and ulna, resulted in a complete
Figure 26. Carnotaurus sastrei. Fourth haemal arch (A-
C) and 10th haemal arch (D-F) in dorsal (A, D), lateral
(B, E), and ventral (C, F) views, he, haemal canal.
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 27
Figure 27. Carnotaurus sastrei. Left scapulo-coracoid
in lateral (A) and medial (B) views. Right clavicle in lateral
(Cl) and medial (C2) views. Sternal plate in lateral (Dl)
and medial {D2) views, ac, acromium; da, subacromial
depression; fc, coracoidal foramen; gl, glenoid cavity; ?,
indeterminate ossification.
change of the original position of the hand. If this
is true, the palmar side moved to a dorsal position,
with digit I assuming a lateral position and digit IV
an internal position.
Table 3. Measurements (in millimeters) of the shoulder
girdle of Carnotaurus sastrei.
Right
Left
Greatest length of scapula and cora¬
coid measured along curve of out¬
side
905
900
Maximum length of scapula
645
610
Maximum breadth of scapular blade
(at level of tubercule for m. triceps)
128
133
Greatest transverse expanse of glenoid
fossa
63
70
Greatest proximodistal expanse of gle¬
noid fossa
80
95
Greatest length of coracoid
430a
430a
Greatest depth of coracoid
290
290
Anteroposterior expanse of sternal
plate
180
—
a = approximate.
Table 4. Measurements (in millimeters) of the forelimb
and manus of Carnotaurus sastrei.
Right
Left
Humerus
Length
285
284
Distal transverse width
86
93
Proximal transverse width
95
100
Transverse width of humeral head
76
75
Anteroposterior width of humeral
head
70
73
Least transverse width of shaft
53
55
Ulna
Length
78
85
Greatest distal transverse width
59
55
Greatest proximal transverse width
50
56
Greatest anteroposterior distance
of proximal end
66
71
Least diameter of shaft
22
26
Radius
Length
73
80
Greatest distal transverse width
63
58
Greatest proximal transverse width
47
48
Least diameter of shaft
30
31
Radiale
Length (proximodistal)
10
—
Transverse width
25
—
Ulnare
Length
18
~
Transverse width
24
Metacarpal I
Length
30
__
Proximal transverse width
23
—
Distal transverse width
23
—
Metacarpal II
Length
37
36
Proximal transverse width
—
Distal transverse width
23
—
Metacarpal III
Length
29
31
Proximal transverse width
—
—
Distal transverse width
—
~
Metacarpal IV
Length
84
—
Pelvic Girdle and Hindlimb
The pelvic bones were articulated to each other,
the sacrum, and the femora. The farmer who found
the specimen, Mr. Sastre, removed some marginal
rock containing the distal part of both ischia. The
length of the ischium (Fig. 31) is, therefore, an es¬
timate, and a few centimeters could possibly be
added between the preserved parts.
Pelvic Girdle (Fig. 31). The ventral bones of the
28 ■ Contributions in Science, Number 416
Bonaparte, Novas, and Coria: Carnotaurus sastrei
10 cm
Figure 28. Carnotaurus sastrei. Right humerus in lat¬
eral (A) and anterior (B) views, dc, deltopectoral crest;
he, humeral head; it, internal tuberosity; rc, radial condyle;
uc, ulnar condyle.
Carnotaurus pelvis are proportionately long, slen¬
der, and straight, and the distal end of the ischium
is expanded into a foot. The acetabulum is large
and open medially, and all three pelvic bones con¬
tribute to it.
The ilium (Figs. 20, 31) is long and low, with the
dorsal border rounded and axially straight, except
its anterior area where it bends ventrad. The ventral
border behind the acetabulum is parallel with the
dorsal border. The iliac blade has a high posterior
border and is concave in its central area. The an¬
terior projection of the ilium is dorsoventrally large,
and it projects ventrad more than in Tyrannosaurus
(Osborn, 1917) or Allosaurus (Madsen, 1976), re¬
sembling the condition of Ceratosaurus (Gilmore,
1920).
The precise union between the pubis and the
ilium is hard to locate. A series of rugosities may
correspond to that union, as tentatively indicated
in Figure 31. The pubic pedicel is wide and short.
The pubis (Fig. 31) has a large contact with the
ischium, and the obturator foramen is largely en¬
circled by bone. The shaft is almost straight, with
a well-developed foot and a long, fused symphysis.
Proximal to the distal foot, both pubes are sepa¬
rated by an elongate aperture.
The ischium (Fig. 31) is long and proportionally
thicker than the pubis. Both ischia are fused for
most of their length. Ventrally, the fused area is
Contributions in Science, Number 416
B
Figure 29. Carnotaurus sastrei. Right radius and ulna
in anterolateral view (A), right radius in medial view (B),
and right radius and ulna with tentative reconstruction of
right manus in palmar view (C). ol, olecranon; op, osseous
process; r, radius; u, ulna; I, II, III, IV, metacarpals I, II,
III, and IV.
present from the distal end up to the middle of the
ventral laminae. Dorsally, the fused area is shorter,
and before the level of the ventral laminae is reached
distad, both ischia begin to separate along their
dorsal edge. Proximally, the ventral laminae are
separated and form a wide acetabular concavity.
The union with the pubis is laminar via a strong
fusion. The posterior pedicel of the ilium shows a
partial fusion with the ischium. In the proximolat-
eral region, almost on its dorsal border, there is an
obvious osseous process for muscular attachment,
possibly for the m. flexor tibialis internus. Accord¬
ing to Gregory and Camp (1918) this is a character
present in Ceratosaurus and Tyrannosaurus, but
not in Allosaurus or Deinonychus.
In the distal area the ischia are intimately fused
and show an anteroposterior, foot-like expansion.
As the distal portions of the ischia are not in clear
contact with the proximal portions, as indicated
above, we do not know the actual length of these
bones.
Flindlimb (Figs. 32, 33). The hindlimbs are rep¬
resented only by the femora and the proximal third
of both tibiae. Unfortunately, erosion destroyed the
rest of both hindlimbs, including the feet. How¬
ever, there is sufficient information from other re-
Bonaparte, Novas, and Coria: Carnotaurus sastreiM 29
I
Figure 30. Carnotaurus sastrei. Right humerus, radius, and ulna in anterior (A), posterior (B), medial (C), and lateral
(D) views. Scale = 10 cm.
cent finds made in Patagonia that permits restora-
tion of the hindlimbs, except for the tarsals and
digits.
In general terms, Carnotaurus had long and slen¬
der hindlimbs, not very different from those of Ba-
hariasaurus (Stromer, 1915). They show both
primitive and derived characters that distinguish
them from Tyrannosaurus and, to a lesser degree,
from Allosaurus.
The left femur (Fig. 32) is reasonably complete,
with the central portion restored after a natural
mold. Its total length is 103 cm, and the average
diameter of the diaphysis is 11 cm, a good indi¬
cation of its slender condition. D. Russell (pers.
comm.) suggests that the size of the femur indicates
that the total weight of this Carnotaurus sastrei
specimen was about 1,350 kg.
The distal end of the femur has a modest trans¬
verse expansion, to 19 cm, and the proximal end
has a length to the major anteromedial axis of 22
cm. The diaphysis is subcylindrical and proximo-
distally convex on the anterior side. In the distal
internal region, a sharp mediodistal crest originates
and ends near the inner tibial condyle. This crest
separates the concave dorsal surface from the me¬
dial side, which is also concave. The inner tibial
condyle projects strongly posteriad, but it is trans¬
versely narrow, whereas the outer condyle is trans¬
versely wider but less pronounced posteriorly. The
major axis of the fibular condyle projects upward
and outward.
At the proximal end the femoral head has a con¬
vex continuity with the major trochanter. The fem-
30 ■ Contributions in Science, Number 416
oral head is, therefore, anteroposteriorly narrow
and not subspheric as in Tyrannosaurus. Nor does
it project mediad as in Tyrannosaurus and Allo¬
saurus, but instead it projects anteromediad. The
lesser trochanter is well developed, but proximally
Figure 31. Carnotaurus sastrei. Pelvis in lateral view.
Ilio-pubic suture suggested by dashed line, ab, anterior
blade; ac, acetabulum; il, ilium; isq, ischium; ob, obturator
foramen; pb, posterior blade; pft, process for the m. flexor
tibialis internus; pub, pubis.
Bonaparte, Novas, and Coria: Carnotaurus sastrei
Table 5. Measurements of pelvic bones and hindlimb
of Carnotaurus sastrei.
Ilium
Greatest length
970
Height above middle of acetabulum
235
Ischium
Greatest expanse from ventral border
of obturator process to posterior
border of iliac pedicel
250
Pubis
Greatest length along anterior
border
880
Greatest length of distal “foot”
252
Greatest length of distal “foot”
106
Greatest transverse width at center of
combined shaft of both pubes
157
Right Left
Femur
Greatest length
—
1030
Greatest diameter of head
—
206
Distal width
197
198
Tibia
Proximal width
120
134
Greatest anteroposterior expanse
245£
i 245a
a = approximate.
it reaches to only slightly below the femoral head.
Both Tyrannosaurus and Allosaurus have dorsally
higher lesser trochanters. The internal, or 4th, tro¬
chanter is not well preserved, but its relative po¬
sition appears to be somewhat higher than in Ty¬
rannosaurus.
The tibiae (Fig. 33A, B) are represented only by
their proximal parts. The articular surface for the
femur is inclined outward and downward. The out¬
er facet is in a posterior position, near the posterior
border of the tibia. The articulation for the fibula
is anteroposteriorly convex. The cnemial crest is
not complete, but it appears to have been well
developed, although perhaps less so than in Allo¬
saurus. The crest for a ligamentous union with the
fibula is dorsally placed, and it lies slightly anterior
to the fibular condyle.
Reconstruction of Tibia-Fibula-Tarsus of Car-
notaurus. Recently, some associated remains of
Carnosauria were described by Martinez et al. (1986).
The material includes incomplete vertebrae and a
complete femur, tibia, fibula, and tarsus, on which
Xenotarsosaurus bonapartei Martinez, Gimenez,
Rodriguez, and Bochatey from the Upper Creta¬
ceous of Patagonia was based.
The similarities of the preserved vertebrae and
limb bones of this theropod with those of Car-
notaurus are so marked that its familial indentifi-
cation is unquestionable, as Martinez et al. (1986)
recognized. The generic differences are based on
Contributions in Science, Number 416
10 cm
Figure 32. Carnotaurus sastrei. Right femur in anterior
(A), lateral (B), posterior (C), and medial (D) views, fc,
fibular condyle; he, femoral head; Idc, mediodistal crest;
It, lesser trochanter; mt, major trochanter; tc, tibial con¬
dyle; tr4, 4th trochanter.
the vertebral morphology, where some differences
exist. But the morphology of the femora strongly
suggests that the hindlimbs of both forms were
probably basically the same.
On this basis we estimate that the tibiae of Car¬
notaurus were slightly shorter than the femora,
possibly about 97 cm in length. It is possible that
the distal region of the tibia had a moderate trans¬
verse expansion and that it might have been fused
to the astragalus and calcaneum, as in Xenotarso¬
saurus, forming a rigid unit. The fibula was prob¬
ably not fused to its neighboring bones, as in Xeno¬
tarsosaurus.
Comparisons of Pelvic Girdle and Hindlimb. The
pelvis of Carnotaurus is typically carnosaurian, with
similarities to Ceratosaurus nasicornis. We see lit¬
tle indication of phylogenetic affinities with Dei-
nonychosauria or Coelurosauria. There are strong
differences in the morphology of the tyrannosaurid
ischia, which lack a distal expansion, and the tyran¬
nosaurid pubis, which is transversely wide and ro¬
bust and lacks an obturator foramen.
The general plan of the sacrum and pelvis is com¬
parable to that of Ceratosaurus, with an axially
extended ilium, an elongated, slender pubis with
an obturator fenestra encircled by bone, and a wide,
laminar contact between the pubis and ischium.
The ischium is elongated and rather slender, and it
has a distal foot-like expansion in both genera
(Marsh, 1896: pi. 10, fig. 1; Gilmore, 1920: 108).
Carnotaurus shows derived characters, relative to
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 31
Figure 33. Carnotaurus sastrei. Proximal end of the
left tibia in lateral (A) and proximal (B) views, compared
with the left tibia, fibula, and tarsus of Xenotarsosaurus
bonapartei in lateral (C) and proximal (D) views, a+c,
astragalus and calcaneum; cn, cnemial crest; e, external
condyle; f, fibula; fc, fibular crest; i, internal condyle; t,
tibia.
Ceratosaurus, in having the dorsal border of the
ilium straight and parallel to the ventral postace-
tabular border, which forms a right angle with the
high posterior border of the ilium, and in having
the pubo-ischiatic symphysis more reduced. We be¬
lieve the similarities are sufficient to consider the
two genera to be closely related.
The hindlimb of Carnotaurus has very slender
proportions, which suggests that it may have been
a more agile form than other carnosaurs, except
for the African forms described as Bahariasaurus
and cf. Spinosaurus (Stromer, 1934: pis. 1,3). How¬
ever, the hindlimb shows significant plesiomorphic
characters relative to Allosaurus and Tyrannosau¬
rus, including a) femoral head projecting antero-
mediad; b) femoral head slightly lower than the
major trochanter, with a convexity between them;
and c) lesser trochanter with a modest dorsal pro¬
jection. These differences suggest that the Allo-
sauridae and Tyrannosauridae are quite distinct from
the Abelisauridae, which includes Carnotaurus.
The Ceratosaurus femur (Gilmore, 1920: fig. 64)
is similar to Carnotaurus in the following features:
a) morphology and relative position of the femoral
head and the major trochanter; b) characters of the
lesser trochanter in its dorsal expression; c) char¬
acters of the distal end, including the relative po¬
sition and morphology of the tibial and fibular con¬
dyles; and d) the position and characters of the
anteromedial crest that separate the medial from
32 ■ Contributions in Science, Number 416
the anterior face of the dorsal side in this part of
the femur.
The similarities with the African Cenomanian
forms (Stromer, 1915, 1931, 1934) are significant,
but some differences are present. The region of the
femoral head-major trochanter is the same, but the
lesser trochanter is more developed. In the distal
end of the femur there are several peculiar char¬
acters in the African forms. In the proximal end of
the tibia, the fibular condyle is not near the pos¬
terior border as in Carnotaurus, but rather it is in
a more anterior position.
Skin
Several fragments of skin impressions were found
underneath the right side of the skeleton. They
correspond to several parts of the body. One frag¬
ment is from the anterior cervical region, and it is
associated with the anterior cervical ribs. Another
fragment is from the scapular area near the glenoid
cavity. Two fragments are from the thoracic region,
and they are associated with the mid-proximal area
of the ribs. The largest available skin impression
corresponds to the lower area of the proximal part
of the tail. There appears to be little variation among
the different fragments. The surface of the skin is
made of rather low, conical protuberances of about
4 to 5 cm in diameter (Fig. 37), each with a modest
keel and separated from one another by about 8
to 10 cm. The surface between the protuberances
is rough, with rather rounded, low, and small gran¬
ules about 5 mm in diameter that are separated
from one another by narrow furrows.
GENERAL COMPARISONS
We consider Carnotaurus sastrei to represent a
new, previously unknown adaptative type within
the Theropoda that has many derived skeletal char¬
acters. These derived characters may give us a wider
perspective of the adaptive potentialities of this
group of dinosaurs, and we believe that the unique
anatomical characters of Carnotaurus must be con¬
sidered in developing the systematics of the Car-
nosauria.
It is necessary, perhaps, to overcome an inter¬
pretation of the 1970’s in which the biogeography
of the Gondwana dinosaurs was only superficially
considered, e.g., Charig (1973: 351): “. .. it may be
that the congeneric dinosaurs found in the various
Gondwanaland regions (if indeed congeneric) were
merely the remnants of populations, once widely
distributed, which have been driven south by var¬
ious agencies into the already isolated peninsulas
of the Southern Hemisphere.” The biogeographic
history of the Cretaceous dinosaurs of Gondwana
was probably completely different from that im¬
plied by Charig’s statement.
Is Carnotaurus sastrei a Carnosaur?
Typical members of the Camosauria are represented
by the Jurassic and Cretaceous families Megalosauri-
Bonaparte, Novas, and Coria: Carnotaurus sastrei
dae, Allosauridae, Ceratosauridae, and Tyranno-
sauridae. In addition, the families Dryptosauridae,
Spinosauridae, and Abelisauridae are very probably
members of the Carnosauria. All of these families
represent a relatively homogeneous group of pred¬
ators of considerable size, in which a progressive
reduction of the forelimbs is documented. These
families are also characterized by long pubes with
a distal foot and with femora slightly longer than
the corresponding tibiae. These characters, in as¬
sociation with other more generalized ones of the
skull, ilium, scapulo-coracoid, and foot, distinguish
the different taxa of the Carnosauria.
The inclusion of Carnotaurus sastrei among the
Carnosauria is fairly conclusive because the out¬
standing reduction of the forelimbs, with the strong
humerus, and the characters of the sacrum, pelvis,
and femur point to a relationship with Ceratosau-
rus. All of these arguments clearly demonstrate that
it is a carnosaur with several derived characters. We
do not believe it is necessary to make comparisons
with other groups of the Theropoda such as Dei-
nonychosauria, Oviraptosauria, Ornithomimosau-
ria, Segnosauria, and Therezinosauridae {fide Rus¬
sell, 1984).
Comparison with Jurassic and Cretaceous
Carnosaurs of South America
All of the significant specimens of South American
carnosaurs are from Argentina. The known remains
from Brasil (Price, 1960) are isolated teeth. Like¬
wise, taxa described by del Corro (1966, 1974) as
Megalosaurus inexpectatus and Megalosaurus
chubutensis lack diagnostic characters and should
be considered as nomina dubia.
Piatnitzkysaurus floresi Bonaparte (1979,1986a)
is an allosaurid from the Middle Jurassic, Callovian,
of Patagonia, showing some plesiomorphic char¬
acters relative to Allosaurus fragilis. The compar¬
ison of Carnotaurus with the Middle Jurassic form
does not provide more information than the com¬
parison with Allosaurus. However, the more prim¬
itive condition of Piatnitzkysaurus suggests that it
is nearer to the common ancestry of Ceratosaur-
idae-Abelisauridae than is Allosaurus.
Genyodectes serus Smith-Woodward, 1901, is
based on the anterior part of a skull and jaws, and
these show significant similarities with Carnotau¬
rus. However, it lacks sufficient diagnostic char¬
acters to identify it with, or differentiate it from,
C. sastrei. In addition, the geographic and strati¬
graphic provenance of Genyodectes serus is so im¬
precise that we are unable to determine if the type
specimen came from Jurassic or Cretaceous de¬
posits, much less from Middle or Upper Cretaceous
deposits. We do not have the necessary stratigraphic
or anatomical data for comparison. There is some
doubt as to the validity of the taxon Genyodectes
serus.
Unquillosaurus ceibalii Powell (1979) is based
on most of a left pubis, bearing a “foot” and an
Figure 34. Carnotaurus 6th cervical vertebra (B, E),
compared with those of Allosaurus (A) and Ceratosaurus
(D), and with the neural arch of the cervical vertebra of
Noasaurus (C, F); and humerus, radius, and ulna (G),
compared with those of Torvosaurus (H), Allosaurus (I),
and Daspletosaurus (J) in lateral view. Not to scale.
anterolateral proximal canal. The obturator fora¬
men is open, which is more derived than in Car¬
notaurus sastrei. Although Unquillosaurus ceibalii
is based on incomplete material, there are sufficient
differences to distinguish it from Carnotaurus.
Xenotarsosaurus bonapartei (Martinez et al.
1986) is from the Bajo Barreal Formation, southern
Chubut Province, Patagonia, probably Senonian in
age. This species is based on two incomplete ver¬
tebrae associated with the femur, tibia, and tarsus
of the right side. Martinez et al. (1986) recognized
some significant differences between Xenotarso¬
saurus and Carnotaurus vertebrae as the depres¬
sions of the vertebral centrae are more marked, the
parapophyses are larger, and the cavities underneath
the prezygapophysis are more developed in Xeno¬
tarsosaurus. These characters suggest Xenotarso¬
saurus is a distinct genus.
A comparison of what is preserved of the hind-
limb of Carnotaurus with the limb bones of Xeno¬
tarsosaurus is very useful, the similarities showing
that both genera are in the same family. Characters
of the femora of both genera are similar, except
that the Carnotaurus femur is relatively longer and
more slender. A comparison of the proximal part
of the tibiae shows common characters in the pos¬
terior position of the external tibial condyle, the
morphology of the cnemial crest, and the relative
position of the fibular crest. Xenotarsosaurus shows
some differences from Carnotaurus in the verte-
Contributions in Science, Number 416
Bonaparte, Novas, and Coria: Carnotaurus sastrei ■ 33
Figure 35. Carnotaurus pelvis (A) compared with that of Ceratosaurus (B) and Tyrannosaurus (C) in lateral view.
Not to scale.
brae, but the characteristics of the hindlimbs show
that both genera belong to the Abelisauridae.
Abelisaurus comahuensis Bonaparte and Novas,
1985, is based on an incomplete skull from the
Allen Formation, Maastrichtian, of Rio Negro
province, Patagonia. Even though the proportions
of the skull are very different from the short, high
skull of Carnotaurus, the two genera share several
diagnostic features at the family level. In the tem¬
poral region of Abelisaurus, plesiomorphic char¬
acters exist in the squamosal and the general shape
of the infratemporal opening. In the orbit, the post¬
orbital almost encloses the orbit ventrally, and the
lachrymal is posteriorly convex, as in Carnotaurus.
The preorbital opening is large, but with different
proportions than that of Carnotaurus. The position
and small size of the maxillary fenestra, which is
slightly separated from the main preorbital opening
but is still within the same general depression, is
similar in both genera. In the supratemporal region
of both genera there are shared characters in the
parietal crest and in the proportions of the open¬
ings. Finally, the ornamentation of the nasals with
rugosities suggests a corneous covering in both gen¬
era. Although Abelisaurus, with its elongated skull,
represents an adaptive type different from Carno¬
taurus, it is evident that they correspond to the
same family.
Noasaurus leali Bonaparte and Powell, 1980, is
a small theropod represented by only a few, but
diagnostic, pieces from the Lecho Formation,
Maastrichtian, of Salta Province, northwestern Ar¬
gentina. The authors recognized a different family,
Noasauridae, which, with some doubt, they re¬
ferred to the Coelurosauria. Considering what is
presently known of the anatomy of Abelisaurus and
Carnotaurus, the original systematic interpretation
may be subject to change.
The maxilla of Noasaurus (Fig. 7H) is relatively
short and high, suggesting that the general shape
of the skull was probably short. The preorbital
opening resembles the condition in Abelisaurus and
Carnotaurus.
34 ■ Contributions in Science, Number 416
The squamosal of Noasaurus (Fig. 71) has a ven¬
tral process without a forward inclination. In this
character it resembles Abelisaurus and Carnotau¬
rus, although there are significant differences in oth¬
er characters of the bone. The quadrate of Noa¬
saurus (Fig. 7G) is long, in agreement with
Abelisaurus and Carnotaurus.
The neural arch of the cervical vertebra of Noa¬
saurus (Fig. 34C, F) shows characters similar to
those of Carnotaurus (Fig. 34B, E). The outstand¬
ing development of the epipophysis and the re¬
duced neural spine, which was assumed to be lack¬
ing by Bonaparte and Powell (1980), correspond to
characters of the peculiar type of cervical neural
arch of Carnotaurus. The cervical ribs of Noasau¬
rus are very different from those of Carnotaurus.
Fundamental characters of the maxilla, squa¬
mosal, and quadrate, and especially the structure
of the cervical neural arch of Noasaurus and Car¬
notaurus, suggest that although the two genera rep¬
resent different families, they may belong to a com¬
mon major taxon, possibly a superfamily.
Comparisons with Tyrannosauridae
The comparisons with this family of Carnosauria
are primarily made with Tyrannosaurus because it
is generally considered very representative of the
family.
The design of the Tyrannosaurus skull (Fig. 7K,
L, M) corresponds to a long, low model, with a
massive lower jaw in lateral view, with a reduced
mandibular fenestra. Carnotaurus has a short, high
skull, and the lower jaw is slender with a large
mandibular fenestra. The significant differences of
the infratemporal opening, orbital and preorbital
region, characters of the jugal, lachrymal, and other
bones have been pointed out in the description.
The general anatomy of the vertebrae posterior
to the axis shows fundamental differences from that
of Tyrannosaurus because of the hypertrophy of
the epipophysis and atrophy of the neural spines in
Carnotaurus, which are derived characters not
Bonaparte, Novas, and Coria: Carnotaurus sastrei
f
Figure 36. Carnotaurus sastrei. Cast of the skeleton at the Museo Argentino de Ciencias Naturales “B. Rivadavia,”
Buenos Aires, mounted in 1986.
present within the Tyrannosauridae. The forelimb
shows derived characters in the outstanding reduc¬
tion of the forearm and the degree of humeral
torsion, characters different from those found in
Tyrannosauridae (Fig. 34J). In the manus, the gen¬
eral plan of the digits is quite different in Carno¬
taurus and tyrannosaurids. In the pelvic girdle (Fig.
35), strong differences are seen in the morphology
of the pubis and ischium, as the obturator foramen
is encircled by bone and the ischium is expanded
distally in Carnotaurus. In the hindlimb, the femur
of Carnotaurus is relatively longer and more slen¬
der, the femoral head is medioanteriorly rather than
medially projected, the lesser trochanter has a lower
position, the 4th trochanter is more dorsally placed,
and the fibular condyle has a different orientation.
The indicated differences between Carnotaurus
and members of Tyrannosauridae make it clear that
Carnotaurus does not belong to this family. The
differences between Tyrannosauridae and Abeli-
sauridae are actually more obvious and better de¬
fined than between the Tyrannosauridae and the
other families of Carnosauria (Allosauridae, Meg-
alosauridae, and Ceratosauridae).
Comparison with Allosauridae
The differences between Allosaurus and Carno¬
taurus are significant and present throughout the
skeleton. In general terms we may consider Allo¬
saurus as a potential ancestor to the Tyrannosaur-
Contributions in Science, Number 416
idae because of several derived characters that per¬
sist or are more fully developed in the Cretaceous
family the anteroventral projection of the
squamosal, the short quadrate, the near absence of
mandibular fenestra, the medial projection of the
femoral head, and the v/ell-developed pubis). None
of these characters are typical of Carnotaurus, in
which the plesiomorphic condition prevails. Car¬
notaurus has several apomorphic characters not
recorded in Allosaurus, including the high, short
skull, frontal horns, a weak contact between the
dentary-splenial and postdentary bones, hypertro¬
phy of epipophyses and atrophy of the neural spines
of the cervicals (Fig. 34), strong fusion and size
reduction of the sacral centra, unique uncinate pro¬
cesses of the anterior caudals, and the extreme re¬
duction of the forearm bones (Fig. 34G, I).
We consider these differences sufficient to dem¬
onstrate that the Allosauridae and Abelisauridae are
separate taxonomic entities, widely separated by
their morphology. They diverged from an evolu¬
tionary stage prior to Allosaurus, perhaps not far
from Piatnitzkysaurus of the Middle Jurassic.
Comparison with Ceratosauridae
The Ceratosauridae is a monotypic family of the
Late Jurassic that, because of its peculiar anatomy,
stimulated the interest of several workers (Marsh,
1884; 1896; Gilmore, 1920; Huene, 1926). C^r^-
tosaurus nasicornis Marsh, from the Morrison
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 35
Formation, Fremont County, Colorado, has several
distinctive characters upon \vhich the family Cer-
atosauridae Marsh was based. According to Gil¬
more (1920), the poor definition of the family was
the main reason for the rejection of the new family
by paleontologists of the time. However, Gilmore
(1920) restudied Ceratosaurus and reaffirmed the
validity of the family.
If, as indicated by Gilmore (1920), Ceratosaurus
has several plesiomorphic characters that identify
it as one of the more primitive post-Trassic The-
ropoda, it is also true that it bears several derived
characters that show specialized features of its anat¬
omy. We do not believe the derived characters of
the sacrum and ilia were ever evaluated in Cera¬
tosaurus, and we think that these characters reveal
specialized features not recorded in contempora¬
neous Carnosauria. The number of vertebrae form¬
ing the synsacrum, the degree of ossification, and
the secondary reduction in size of the five fused
centra represent a suite of derived characters not
recorded in other Jurassic or Cretaceous Carnosau¬
ria of the Northern Hemisphere. The synsacrum of
Ceratosaurus includes two anterior vertebrae that
serve a sacral function (“presacrals 22 and 23”) and
a posterior sacral, indicated as “caudal 1 ” (Gilmore,
1920: pi. 21). Such a specialized sacrum fits between
the extended ilia of Ceratosaurus, which are rel¬
atively more developed and derived than in Allo-
saurus. The ratio between the lengths of the radius
and femur of Ceratosaurus and Allosaurus also
illustrates the degree of limb disparity; the ratio is
more derived in Ceratosaurus than in Allosaurus.
The radius of Ceratosaurus is 24% the length of
the femur, whereas in Allosaurus (USNM 4734) it
is 28%. This may be a good indication, when con¬
sidered with the other characters cited above, that
Ceratosaurus was not the most primitive of the
post-Triassic theropods, but an adaptive type char¬
acterized by a mix of plesiomorphic and apomor-
phic characters that followed a different evolution¬
ary path than did the Allosauridae-Tyrannosauridae
line.
In spite of the different ages of Ceratosaurus
(Late Jurassic) and of Carnotaurus (Middle or Late
Cretaceous) and the presence in both genera of very
different characters, we believe that they share a
common ancestor. Both represent a comparable
adaptive level based on the plesiomorphic and apo-
morphic characters they share.
We shall consider first the similarities and then
the differences. In addition to the skulls being sim¬
ilar in the height of the premaxilla in the narial
region, they have: a small maxillary fenestra located
near the preorbital opening; a jugal that lacks an
anterior projection overlying the maxilla and which
does not border the preorbital opening; a large in¬
fratemporal opening, without an anterior projec¬
tion of the quadratojugal and squamosal; a dor-
soventrally long quadrate; and a squamosal with a
slender, rod-like ventral projection, not oriented
anteroventrally.
36 ■ Contributions in Science, Number 416
There are many similarities in the postcranial
skeleton: the centra are well fused in the sacrum
and secondarily reduced; the pelvis (Fig. 35A, B) is
axially elongated; the pubis has the obturator fo¬
ramen surrounded by bone; there is a large laminar
contact between pubis and ischium; the ischium
has a distal expansion; the femoral head is lower
than the major trochanter and directed anterome-
diad; the lesser trochanter is poorly developed dor-
sally; and digits of the manus have short phalanges.
The more significant differences between the genera
include three premaxillary teeth in Ceratosaurus
and four in Carnotaurus; an absence of nasal horn
or lachrymal protuberances in Carnotaurus; an ab¬
sence of frontal horns in Ceratosaurus; skull short
and high in Carnotaurus; lachrymal posteriorly
convex in anterior projection of post¬
orbital absent in Ceratosaurus; a very wide jugal
in Carnotaurus; more developed mandibular fe-
nestrae in Carnotaurus; cervical vertebrae that are
basically different, being much more derived in Car¬
notaurus (Fig. 34D, B); anterior caudal vertebrae
more derived in Carnotaurus; and major differ¬
ences in the humerus (Galton and Jensen, 1979: fig.
3T) and in the forearm bones.
Some of these differences rule out the possibility
that Ceratosaurus was in an ancestral position to
Carnotaurus. These include the nasal and lachry¬
mal protuberances and the presence of only three
teeth in the premaxilla. Other important characters,
such as the peculiar morphology of the cervical and
caudal vertebrae of Carnotaurus, do not have any
morphological similarities in Ceratosaurus, pre¬
venting us from relating these two genera directly.
Finally, the similarities, even ignoring those that
are plesiomorphic characters, are broad enough to
suggest a significant, close phylogenetic relationship
between Ceratosaurus and Carnotaurus. At pres¬
ent it is not easy to define such a relationship, but
it is obviously closer than with any other family of
Carnosauria. This means that in spite of several
good differences at the family level, Ceratosauridae
and Abelisauridae appear to be related, and they
may belong in the same superfamily. Such an ar¬
rangement may be useful to unite them and, at the
same time, distinguish them from the remaining
families of Carnosauria.
Comparison with Theropods from
Egypt and India
The theropod remains from the Cenomanian of
Baharija, Egypt, include significant remains of Car¬
nosauria. They correspond to Spinosaurus aegyp-
tiacus Stromer, 1915, Carcharodontosaurus sa-
haricus Stromer, 1931, and Bahariasaurus ingens
Stromer, 1934. According to Stromer, these three
genera are characterized by elongate, slender ap¬
pendicular bones, without large expansions in the
distal end of the femur. In the case of Baharia¬
saurus, there are plesiomorphic characters in the
proximal region of the femur, with the femoral
Bonaparte, Novas, and Coria: Carnotaurus sastrei
head projecting medioanteriad and lower than the
major trochanter.
The possibility of comparison between the Af¬
rican forms and the Abelisauridae (Abelisaurus,
Contributions in Science, Number 416
CarnotauruSy and Xenotarsosaurus) are limited be¬
cause of the fragmentary condition of the former,
by the notable derived characters of the Spinosau-
rus vertebrae, and in part, too, because the African
Bonaparte, Novas, and Coria: Carnotaurus sastreiM 37
Figure 38. Carnotaurus sastrei. Reconstruction of the skeleton.
specimens were damaged or destroyed during World
War II. But, in spite of the limitations, there are
anatomical features that suggest more similarities
between Carnotaurus (Abelisauridae) and the Af¬
rican forms than between Abelisauridae and Ty-
rannosauridae. For example, there are indications
that the basic characters of the hindlimb of Car¬
notaurus and the African forms are very similar
and differ markedly from those of the Tyranno-
sauridae. Characters of the latter include a robust
femora, with the proximal head projecting mediad
and higher than the major trochanter and with the
lesser trochanter more dorsally developed.
We do not have sufficient information to evaluate
the validity of the Spinosauridae; however, the ver¬
tebral characters and the features of the lower jaw
and teeth appear to be sufficiently derived to rec¬
ognize it as valid.
From a paleobiogeographic approach, it has been
considered that the Spinosauridae and Abelisaur¬
idae would be endemic families for the Cretaceous
of Gondwana (Bonaparte, 1986b), whereas the Ty-
rannosauridae appear to be endemic to Laurasia. If
this interpretation, based on paleogeographic events
of great magnitude, such as the separation of Lau¬
rasia and Gondwana before the end of the Jurassic,
is correct, we can hypothesize that the few simi¬
larities cited between the African and South Amer¬
ican Middle and Late Cretaceous Carnosauria result
from a common biogeographic history.
The carnosaurs from the Upper Cretaceous of
India, Indosaurus mattleyi (Huene and Mattley,
1933), and Indosuchus raptorius (Fluene and Matt-
ley, 1933), were reviewed by Walker (1964) and
by Chatterjee (1978). The latter described new ma¬
terial of Indosuchus stored at the AMNH. Chat¬
terjee agreed with Walker’s interpretation that In¬
dosaurus is a megalosaurid (=Allosauridae) and that
Indosuchus is a tyrannosaurid. The assignment of
38 ■ Contributions in Science, Number 416
Indosuchus to the Tyrannosauridae appears with¬
out basis, primarily because the maxilla of Indo¬
suchus lacks the typical maxillary fenestra of the
tyrannosaurids.
Our interpretation of Indosuchus is that it rep¬
resents a different family than the Tyrannosauridae,
possibly corresponding to the Abelisauridae, in
which the maxillary fenestra exists in a reduced
form within the main preorbital vacuity.
We do not have enough information for a full
systematic evaluation of Indosaurus; however, we
call attention to the imprecise suggestion that it is
a “megalosaur” (Chatterjee, 1978:573), which means
less than to say “carnosaur.” This assignment was
based on fragments discovered in latitudes very dis¬
tant from where Jurassic megalosaurs are recorded.
Lacking the necessary evidence, we believe it is
better to suggest that Indosaurus may represent a
new, unknown family rather than force geography
and chronology by identifying it with European or
North American forms.
Summary
Comparisons of Carnotaurus sastrei with other
species of Carnosauria allow the following inter¬
pretations:
1. The cranial characters of Carnotaurus sastrei
are good indicators of familial affinities with Abeli-
saurus comahuensis, type species of the family
Abelisauridae Bonaparte and Novas, 1985. This
permits reference, without doubt, of C. sastrei to
that family of Carnosauria.
2. As indicated in the description and in com¬
parisons with Cretaceous theropods of Laurasia,
the Abelisauridae have a suite of characters that
distinguish them from the several theropod families
of that supercontinent, especially the Tyranno¬
sauridae.
3. The phylogenetic relationship between the
Bonaparte, Novas, and Coria: Carnotaurus sastrei
Figure 39. Carnotaurus sastrei. Model by Stephen Czerkas.
Abelisauridae and Noasauridae, based on the mor¬
phology of the maxilla, quadrate, squamosal, and
cervical neural arch, appear convincing in view of
the following synapomorphies: significant reduc¬
tion of the additional preorbital opening, reduction
of the cervical neural spines, strong anteriad de¬
velopment of the epipophyses, and the lateral crest
from the epipophyses. We interpret these characters
as suggesting a monophyletic relationship for the
two families.
4. Theropods from the Cenomanian of Egypt
resemble the Abelisauridae more than the Tyran-
nosauridae, at least in the hindlimb organization.
5. Comparisons between C. sastrei and Jurassic
theropods suggest that the Abelisauridae, and prob¬
ably Noasauridae, may by phylogenetically related
to the Ceratosauridae, but we cannot clearly define
this relationship at present. The phylogenetic re¬
lationship may be one of common ancestry at the
familial level, based on the common possession of
several apomorphic characters, such as sacrum with
several fused vertebrae with reduced centrae,
anteroposteriorly elongated ilium with extended
supra-acetabular crest, the ischia distally expanded
as a “foot,” and a very slender pubis shaft. Several
Contributions in Science, Number 416
plesiomorphic characters, e.g., a large infratemporal
opening, manus with four digits, obturator foramen
enclosed by bone, poor development of the lesser
trochanter, femoral head medioanteriorly pro¬
jected, have also been noted.
DISCUSSION
The comparisons of Carnotaurus sastrei with other
theropods have permitted us to recognize similar¬
ities and differences at different levels, with taxa
from South America and other continents.
The phylogenetic relationship between the Al-
losauridae and Tyrannosauridae that is recognized
by some authors (Paul, 1984; Gauthier, 1986) ap¬
pears well based, and it seems to represent one of
the main lineages of the Carnosauria. Another lin¬
eage, of similar importance, is formed by the Cer-
atosauridae-Abelisauridae-Noasauridae. We hy¬
pothesize that from Jurassic ancestors of Pangean
distribution two vicariant groups developed, lead¬
ing to the Cretaceous families Tyrannosauridae in
Laurasia and Abelisauridae in Gondwana.
Following the results of the comparisons detailed
above, we believe it is convenient to recognize a
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 39
systematic entity at the superfamily level—the Cer-
atosauroidea—to contain the Ceratosauridae,
Abelisauridae, and Noasauridae, whose main ra¬
diation and Cretaceous evolution was in Gon-
dwana.
The systematic interpretation of this hypothesis
may be stated as:
Superfamily Tyrannosauroidea
Family Allosauridae
Family Tyrannosauridae
Superfamily Ceratosauroidea
Family Ceratosauridae
Family Abelisauridae
Family Noasauridae
This interpretation presents us with the novel
opportunity of interpreting as carnosaurs a family—
the Noasauridae—to date known only from small¬
sized species such as Noasaurus leali, which has a
skull approximately 10 to 12 cm long. The inter¬
pretation is based on the synapomorphies of the
maxilla and peculiar characters of the cervical neu¬
ral arch. The presence of Noasauridae among the
Carnosauria, an infraorder normally characterized
by large predators, may relate to the long isolation
of the Gondwana continents through the Creta¬
ceous (Bonaparte, 1986b) and the apparent absence
of Coelurosauria in the Cretaceous of Gondwana.
The Cretaceous isolation of both supercontinents
would have provided an opportunity for the Gon-
dwanian Carnosauria to occupy the ecological niches
of small predators.
ACKNOWLEDGMENTS
We thank the Sastre family for their cooperation during
the fieldwork. The National Geographic Society provided
financial support for the fieldwork. The Consejo Nacional
de Investigaciones Cientificas y Tecnicas and authorities
of the Museo Argentino de Ciencias Naturales supported
most of the laboratory work. S. and S. Czerkas helped in
several aspects of our work. D. Russell, P. Currie, and J.
McIntosh critically reviewed the manuscript, but the au¬
thors remain solely responsible for all aspects of this work.
Special thanks go to C.C. Black and K.E. Campbell for
their help in publishing this paper.
LITERATURE CITED
Barsbold, R. 1974. Saurornithoididae, a new family of
small theropod dinosaurs from central Asia and
North America. In Results of the Polish-Mongolian
Palaeontological Expedition, Part V, Z. Kielan-Ja-
worowska, ed. Palaeontologia Polonica, 30:5-22.
Bonaparte, J.F. 1979. Dinosaurs: a Jurassic assemblage
from Patagonia. Science, 205:1377-1379.
-. 1985. A horned Cretaceous carnosaur from Pat¬
agonia. National Geographic Research, 1:149-151.
-. 1986a Les dinosaures (carnosaures, allosaur-
ides, sauropodes, cetiosaurides) du Jurassique Moy-
en de Cerro Condor (Chubut, Argentine). Annales
de Paleontologie (Vertebres-Invertebres), 72(3 et 4):
247-386.
-. 1986b. Flistory of the terrestrial Cretaceous
vertebrates of Gondwana. IV Congreso Argentino
40 ■ Contributions in Science, Number 416
de Paleontologia y Bioestratigrafia, Mendoza, Ar¬
gentina, 2:63-95.
Bonaparte,]., and F.E. Novas. 1985. Abelisaurus coma-
huensisy n. gen., n. sp., Carnosauria del Cretacico
tardio de Patagonia. Ameghiniana, 21(2-4):259-265.
Bonaparte, ]., and J.E. Powell. 1980. A continental as¬
semblage of tetrapods from the Upper Cretaceous
beds of El Brete, northwestern Argentina (Sauro-
poda-Coelurosauria-Carnosauria-Aves). Memories
de la Societe Geologique de France, N.S., 139:19-
28.
Charig, A. 1973. Jurassic and Cretaceous dinosaurs. In
Atlas of Paleobiogeography, A. Hallam, ed. London,
Elsevier Press, pp. 339-352.
Chatterjee, S. 1978. Indosuchus and Indosaurus, Cre¬
taceous carnosaurs from India. Journal of Paleon¬
tology, 52(3):570~580.
Colbert, E.H., and D.A. Russell. 1969. The small Cre¬
taceous dinosaur Dromaeosaurus. American Mu¬
seum Novitates, 2380:1-49.
Currie, P.J. 1987. Bird-like characteristics of the jaws
and teeth of troodontid theropods (Dinosauria-Saur-
ischia). Journal of Vertebrate Paleontology, 7:72-
81.
del Corro, G. 1966. Un nuevo dinosaurio carnivoro del
Chubut. Museo Argentino de Ciencias Naturales “B.
Rivadavia,” Comunicaciones Paleontologia, 1(1):1-
14.
-. 1974. Un nuevo dinosaurio megalosaurio (car-
nosaurio) del Cretacico de Chubut (Argentina). Mu¬
seo Argentino de Ciencias Naturales “B. Rivadavia,”
Comunicaciones Paleontologia, l(5):37-44.
Gabon, P.M., and J. Jensen. 1979. A new large theropod
dinosaur from the Upper Jurassic of Colorado. Brig¬
ham Young University, Geology Studies, 26(2):l-2.
Gauthier, J.A. 1986. Saurischian monophyly and the
origin of birds. In The Origin of Birds and the Evo¬
lution of Flight, K. Padian, ed. Memoirs of the Cal¬
ifornia Academy of Science, 8:1-56.
Gilmore, C.W. 1920. Osteology of the carnivorous Di-
nosauria in the United States National Museum.
United States National Museum, Bulletin, 110:1-
159.
Gregory. W.K., and C.L. Camp. 1918. Studies in com¬
parative myology and osteology. American Museum
of Natural History, Bulletin, 3rd Part, 2(38):447-
564.
Huene, F. von. 1926. The carnivorous Saurischia in the
Jura and Cretaceous formations principally in Eu¬
rope. Revista del Museo de La Plata, Universidad
Nacional de La Plata, Argentina, 29:35-167.
Huene, F. von, and C.A. Mattley. 1933. The Cretaceous
Saurischia and Ornithischia of the central provinces
of India. Paleontologia Indica, 21(l):l-74.
Lambe, L.M. 1917. The Cretaceous theropodous di¬
nosaur Gorgosaurus. Memoirs of the Geological
Survey of Canada, 100:1-84.
Madsen, J.H. 1976. Allosaurus fragilis: a revised os¬
teology. Utah Geological and Mineral Survey, Bul¬
letin, 109:1-163.
Maleev, E.A. 1955. Carnivorous dinosaurs of Mongolia.
Priroda, 6:112-115. [In Russian]
Marsh, O.C. 1884. Principal characters of American
Jurassic dinosaurs, 8: The order of Theropoda.
American Journal of Science, 27:329-340.
--—. 1896. The dinosaurs of North America. United
States Geological Survey, 16th Annual Report, (1):
133-244.
Martinez, R., O. Gimenez, J. Rodriguez, and G. Bochatey.
Bonaparte, Novas, and Coria: Carnotaurus sastrei
1986. Xenotarsosaurus bonapartei n. gen., n. sp.
(Carnosauria, Abelisauridae), un nuevo Theropoda
de la Fm. Bajo Barreal, Chubut, Argentina. IV Con-
greso Argentine de Paleontologia y Bioestratigrafia,
Mendoza, Argentina, 2:23-31.
Osborn, H.F. 1912. Crania of Tyrannosaurus and Al-
losaurus. American Museum of Natural History,
Memoirs (n.s.), 1:1-30.
-. 1917. Skeletal adaptations of Ornitholestes,
Strutiomimus and Tyrannosaurus. American Mu¬
seum of Natural History, Bulletin, 35:733-771.
Osmolska, H. 1979. Nasal salt gland in dinosaurs. Acta
Paleontologia Polonica, 24(2):205-215.
-. 1982. Hulsatnpes perlei n. g., n. sp. (Deinony-
chosauria, Saurischia, Dinosauria) from the Upper
Cretaceous Barum Goyot Formation of Mongolia.
Neue Jahrbuch fiir Geologic und Palaeontologie, 7:
440-448.
Ostrom, J.H. 1961. Cranial morphology of the hadro-
saurian dinosaurs of North America. American Mu¬
seum of Natural History, Bulletin, 122(2):3-196.
-. 1969. OsxtoXo^ oi Deinonychus antirrhopus,
an unusual theropod from the Lower Cretaceous of
Montana. Bulletin, Peabody Museum, 30:1-165.
Paul, G. 1984. The segnosaurian dinosaurs: relics of the
prosauropod-ornithischian transition? Journal of
Vertebrate Paleontology, 4(4):507-515.
Powell, J.E. 1979. Sobre una asociadon de dinosaurios
y otras evidencias de vertebrados del Cretacico su¬
perior de la region de La Gandelaria, Provincia de
Salta, Argentina, Ameghiniana, 16(l-2):191-204.
Price, L. 1960. Dentes de theropoda num testemunho
de Sonda no Estado do Amazonas. Anais da Aca¬
demia Brasileira de Ciencias, 32(l):79-84.
Contributions in Science, Number 416
Romer, A.S., and T. Parsons. 1978. The Vertebrate Body.
Shorter Version. W.B. Saunders Co., Philadelphia,
pp. 1-476.
Russell, D.A. 1970. Tyrannosaurs from the Late Cre¬
taceous of western Canada. National Museum of
Canada, Publications in Paleontology, 1:1-34.
-. 1984. A check list of the families and genera
of North American dinosaurs. National Museums
of Canada, Syllogeus, 53:1-35.
Smith-Woodward, A. 1901. On some extinct reptiles
from Patagonia of the genera Miolania, Dinilysia
and Geniodectes. Proceedings for the General Meet¬
ings for the Scientific Business of the Zoological
Society of London, 71:169-184.
Stromer, E. 1915. Das original der Theropoden Spi-
nosaurus aegyptiacus n. gen., n. sp. Abhandlungen
der Bayerischen Akademie der Wissenschaften, 28(3):
1-32.
-. 1931. Ein skellett-Rest von Carcharodonto-
saurus. Abhandlungen der Bayerischen Akademie
der Wissenschaften, N.F., 9:1-23.
-. 1934. Ergebnisse der Forschungsreisen Prof. E.
Stromers in den Wusten Agyptens. II. Wirbeltierreste
der Baharije-Stufe (unterstes Cenoman). 13. Dino¬
sauria. Abhandlungen der Bayerischen Akademie der
Wissenschaften, N.F., 22:5-79.
Walker, A.D. 1964. Triassic reptiles from the Elgin area.
Ornithosuchus and the origin of carnosaurs. Philo¬
sophical Transactions of the Royal Society of Lon¬
don, (B), 248(744):53-134.
Submitted 9 January 1987; accepted 17 July 1989.
Bonaparte, Novas, and Coria: Carnotaurus sastrei M 41
Live Music
Archive
Librivox
Free Audio
Metropolitan Museum
Cleveland
Museum of Art
Internet
Arcade
Console Living Room
Open Library
American
Libraries
TV News
Understanding
9/11